--- Log opened Sun Nov 28 00:00:08 2010 | ||
-!- elmom [~elmom@hoasnet-fe29dd00-137.dhcp.inet.fi] has joined #hplusroadmap | 00:12 | |
-!- joshcryer [~g@unaffiliated/joshcryer] has joined #hplusroadmap | 00:12 | |
-!- Juul [~Juul@static.88-198-13-205.clients.your-server.de] has joined #hplusroadmap | 01:18 | |
-!- bdesk_ [~argriffi@brcdesk18.statgen.ncsu.edu] has joined #hplusroadmap | 01:57 | |
-!- augur [~augur@c-71-196-120-234.hsd1.fl.comcast.net] has quit [Ping timeout: 240 seconds] | 01:57 | |
-!- superkuh_ [hukrepus@unaffiliated/superkuh] has joined #hplusroadmap | 01:57 | |
-!- joshcryer [~g@unaffiliated/joshcryer] has quit [Ping timeout: 272 seconds] | 01:58 | |
-!- superkuh [hukrepus@unaffiliated/superkuh] has quit [Ping timeout: 240 seconds] | 02:01 | |
-!- bdesk [~argriffi@unaffiliated/bdesk] has quit [Ping timeout: 240 seconds] | 02:01 | |
-!- augur [~augur@c-71-196-120-234.hsd1.fl.comcast.net] has joined #hplusroadmap | 02:01 | |
-!- joshcryer [~g@c-24-8-138-2.hsd1.co.comcast.net] has joined #hplusroadmap | 02:01 | |
-!- joshcryer [~g@c-24-8-138-2.hsd1.co.comcast.net] has quit [Changing host] | 02:01 | |
-!- joshcryer [~g@unaffiliated/joshcryer] has joined #hplusroadmap | 02:01 | |
-!- Netsplit *.net <-> *.split quits: fitzsim | 02:07 | |
-!- elmom [~elmom@hoasnet-fe29dd00-137.dhcp.inet.fi] has quit [Ping timeout: 264 seconds] | 02:08 | |
-!- elmom [~elmom@hoasnet-fe29dd00-137.dhcp.inet.fi] has joined #hplusroadmap | 02:09 | |
-!- Juul [~Juul@static.88-198-13-205.clients.your-server.de] has quit [Ping timeout: 260 seconds] | 02:39 | |
-!- Redeemer [~lorddeeme@c-75-72-218-226.hsd1.mn.comcast.net] has joined #hplusroadmap | 02:40 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has quit [Ping timeout: 245 seconds] | 02:51 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has joined #hplusroadmap | 02:56 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has quit [Ping timeout: 245 seconds] | 03:59 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has joined #hplusroadmap | 04:04 | |
-!- klafka [~klafka@cpe-74-74-152-155.rochester.res.rr.com] has quit [Quit: This computer has gone to sleep] | 04:57 | |
-!- killall-9 [~paulc@diana.null.ro] has joined #hplusroadmap | 05:56 | |
-!- codeshepherd [~Deepan@122.167.76.104] has joined #hplusroadmap | 06:58 | |
-!- codeshepherd [~Deepan@122.167.76.104] has quit [Client Quit] | 07:03 | |
-!- fitzsim [~fitzsim@bas3-toronto06-2925098524.dsl.bell.ca] has joined #hplusroadmap | 07:10 | |
kanzure | fitzsim: the git repo looks much better | 07:50 |
---|---|---|
kanzure | what do you suggest? should i just overwrite the public nanoengineer.git and tell everyone they should clone it again? | 07:50 |
kanzure | hi bdesk_ | 07:53 |
fitzsim | kanzure: do you have a URL I can test? | 08:00 |
kanzure | i can make this public but you'll have to reset your remote/origin later. one sec | 08:00 |
kanzure | http://diyhpl.us/cgit/nanoengineer-fixed/ | 08:03 |
kanzure | cgit doesn't seem to deal with branches at all? it only shows "master" for me | 08:04 |
kanzure | i guess git push hasn't finished yet | 08:04 |
fitzsim | ok | 08:05 |
kanzure | git branch -a lists all branches as remotes/cad and remotes/whatever | 08:05 |
kanzure | maybe that's part of the problem? (i've never worked with remote branches) | 08:05 |
kanzure | git push is now done. | 08:05 |
fitzsim | email addresses show up in commit messages on the web. Is that OK? | 08:06 |
kanzure | i think there's a cgit option +enable-remote-branches | 08:06 |
-!- augur [~augur@c-71-196-120-234.hsd1.fl.comcast.net] has quit [Ping timeout: 240 seconds] | 08:06 | |
kanzure | fitzsim: yeah it's a fucken' git repo | 08:06 |
fitzsim | kanzure: re: remote branches, see item 4 here: http://pauldowman.com/2008/07/26/how-to-convert-from-subversion-to-git/ | 08:09 |
fitzsim | cgit 0.8.3 has a noplainemail option to hide email addresses on the web | 08:14 |
fitzsim | kanzure: no need for swearing, btw :-) | 08:14 |
-!- mheld [~mheld@pool-173-76-224-45.bstnma.fios.verizon.net] has joined #hplusroadmap | 08:16 | |
fitzsim | kanzure: testing a "time git clone" of the test repo | 08:21 |
fitzsim | real1m7.427s | 08:23 |
fitzsim | user0m16.061s | 08:23 |
fitzsim | sys0m4.681s | 08:23 |
fitzsim | 08:23 | |
fitzsim | Receiving objects: 100% (122496/122496), 70.61 MiB | 1.52 MiB/s, done. | 08:24 |
fitzsim | $ git branch -r | 08:26 |
fitzsim | origin/HEAD -> origin/master | 08:26 |
fitzsim | origin/master | 08:26 |
kanzure | do you have those numbers from last time? | 08:26 |
kanzure | try git branch -a | 08:26 |
fitzsim | $ git branch -a | 08:26 |
fitzsim | * master | 08:26 |
fitzsim | remotes/origin/HEAD -> origin/master | 08:26 |
fitzsim | remotes/origin/master | 08:26 |
kanzure | :( | 08:26 |
kanzure | "So you'll need to convert them manually (or write a script to do it if you have a lot, I'll leave that as an exercise for the reader)." | 08:26 |
kanzure | okie dokie | 08:26 |
kanzure | looks like scripting it is :D | 08:27 |
fitzsim | sounds fun :) | 08:27 |
kanzure | but 1min 7sec cloning time makes me much happier | 08:27 |
kanzure | although it looks like you might be on a reasonably modern connection? | 08:27 |
-!- uniqanomaly [~ua@dynamic-78-8-83-20.ssp.dialog.net.pl] has quit [Remote host closed the connection] | 08:29 | |
fitzsim | yeah, I'm on 16Mbps | 08:29 |
-!- uniqanomaly [~ua@dynamic-78-8-83-20.ssp.dialog.net.pl] has joined #hplusroadmap | 08:29 | |
fitzsim | I like that git status is so much faster | 08:29 |
fitzsim | I think that shouldn't be affected by there being more branches | 08:30 |
fitzsim | kanzure: if you want to try converting one remote tag to a branch and pushing, I can try pulling it | 08:30 |
kanzure | ok one sec | 08:31 |
kanzure | star wars rogue squadron will have to wait :3 | 08:31 |
-!- killall-9 [~paulc@diana.null.ro] has quit [Quit: Leaving] | 08:31 | |
kanzure | some of the tagnames are "tags/cad" | 08:32 |
kanzure | i guess the new name can be "cad" | 08:32 |
kanzure | it's just kind of awkward saying git tag tags/cad tags/cad | 08:33 |
fitzsim | yeah, good idea | 08:34 |
kanzure | there's also a "Distribution" and "tags/Distribution" listed by git branch -r so i don't feel safe doing git tag Distribution tags/distribution | 08:34 |
kanzure | oh -r remotes. righto | 08:34 |
kanzure | http://diyhpl.us/~bryan/irc/nanoengineer/branches.txt | 08:35 |
fitzsim | the other thing is where did the branches from the SVN repository end up? | 08:36 |
fitzsim | shouldn't they have been converted to git branches? | 08:36 |
kanzure | the branches are the ones that are plain-listed | 08:36 |
kanzure | http://diyhpl.us/cgit/nanoengineer/tree/branches | 08:36 |
kanzure | Distribution, cad, sim, wware_jni_nd1 | 08:36 |
kanzure | everything else is prefixed with "tags/" http://diyhpl.us/~bryan/irc/nanoengineer/branches.txt | 08:37 |
fitzsim | right, so I'd expect them to show up in the "branch -a" output | 08:37 |
kanzure | on my copy of nanoengineer-fixed.git they are listed as remote branches | 08:37 |
fitzsim | did you pass -b branches on the git-svn command line? | 08:38 |
kanzure | yes | 08:38 |
kanzure | svn clone https://polosims-svn.svn.cvsdude.com/polosims/ --no-metadata -A authors.txt -t tags -b branches -T trunk nanoengineer | 08:38 |
kanzure | uh, git svn clone | 08:39 |
fitzsim | and the plain-listed names all appear as directories in the original svn directory? | 08:39 |
kanzure | well, i haven't done a vanilla svn co yet | 08:39 |
fitzsim | is there a ViewCVS for the SVN repo? | 08:39 |
kanzure | mostly because it will be 10 hours until it's done or something | 08:39 |
kanzure | yes.. one sec | 08:40 |
kanzure | https://polosims-svn.browse.cvsdude.com/polosims | 08:41 |
fitzsim | Username/password? | 08:42 |
kanzure | uh | 08:42 |
kanzure | i'd rather not | 08:42 |
fitzsim | heh, thought there may be an anonymous one :) | 08:42 |
kanzure | basically, ViewCVS' root folder looks exactly like http://diyhpl.us/cgit/nanoengineer/tree/ | 08:42 |
fitzsim | ok | 08:43 |
kanzure | (except there's a weird Mark Sims email address html-encoding-error on the page that messes up the html at the footer) | 08:43 |
kanzure | yeah it doesn't help that some of the tags and branches have the same name | 08:44 |
kanzure | only the "wware_jni_nd1" branch is unique in its name | 08:44 |
fitzsim | yeah, so maybe we do need the "tags/" namespace | 08:44 |
fitzsim | I'm new enough to git that I don't know the convention here | 08:45 |
kanzure | well, i'll do wware_jni_nd1 for you | 08:45 |
fitzsim | also, it seems strange to me that git-svn doesn't make the branches in branches/ into first-class git branches automatically | 08:45 |
kanzure | i ran git tag Distribution tags/Distribution before i realized the namespace issues | 08:47 |
kanzure | so presumably i should undo that | 08:47 |
fitzsim | yeah | 08:48 |
fitzsim | make sure you back-up the current repo before experimenting | 08:48 |
fitzsim | so you won't have to re-clone for another 14 hours | 08:48 |
fitzsim | ah, I guess this is a consequence of git-svn handling ongoing interaction with the SVN server | 08:53 |
kanzure | fix fix fix | 08:54 |
kanzure | i want: git tag Distribution refs/remotes/tags/Distribution; git branch -r -d tags/Distribution | 08:56 |
fitzsim | http://stackoverflow.com/questions/2006265/is-there-an-standard-naming-convention-for-git-tags | 09:01 |
kanzure | is editing nanoengineer's namespace a good idea? there's probably shitloads of references to branches and tags on the wiki | 09:02 |
fitzsim | good question | 09:03 |
fitzsim | this could be an opportunity to improve the naming | 09:03 |
kanzure | eek tags/release-1.0.1_2008-05-06_0815-EST | 09:04 |
fitzsim | the standard git convention seems to be "1.1" for the 1.1 maintenance branch and "v1.1" for the 1.1 release | 09:04 |
fitzsim | yeah | 09:04 |
fitzsim | so that would become: v1.0.1_2008-05-06_0815-EST | 09:04 |
kanzure | short story of mine.. i used to be more into perl scripting before python | 09:05 |
kanzure | one day in the lab i found myself writing filenames like 2008-09-14.pl | 09:05 |
-!- augur [~augur@74.174.236.120] has joined #hplusroadmap | 09:05 | |
kanzure | it was after looking back over hundreds of files named by date that i realized how awful everything was | 09:05 |
kanzure | hm so you suggest keep the date in the tagname | 09:05 |
fitzsim | release_candidate_1-1.1.0_2008-06-04_1000-EST would become: v1-1.1.0_2008-06-04_1000-ESTrc | 09:06 |
fitzsim | v1-1.1.0_2008-06-04_1000-EST-rc | 09:06 |
fitzsim | or: v1-1.1.0_2008-06-04_1000-EST-candidate | 09:06 |
kanzure | huh? why ^v1-? | 09:07 |
fitzsim | I'm looking at: http://diyhpl.us/cgit/nanoengineer/tree/tags | 09:07 |
fitzsim | ah, I see, yeah | 09:07 |
kanzure | oh 1- was already in the name. got it. | 09:07 |
fitzsim | rc1 | 09:07 |
-!- NOMADICUS [~kvirc@DSLPool-net209-116.wctc.net] has quit [Quit: KVIrc 4.0.2 Insomnia http://www.kvirc.net/] | 09:08 | |
fitzsim | v1.1.0_2008-06-04_1000-EST-release_candidate_1 | 09:08 |
fitzsim | vcad, vsim? | 09:08 |
kanzure | i don't know why these are here anyway: | 09:09 |
kanzure | tags/release-1.0.1_2008-05-06_0815-EST | 09:09 |
kanzure | tags/release-1.0.1_2008-05-06_1200-EST | 09:09 |
kanzure | basically "a few hours later we decided this was going to be a new tag" | 09:09 |
fitzsim | yeah | 09:09 |
fitzsim | that's fine though | 09:09 |
fitzsim | we could even tack on -release | 09:10 |
kanzure | git diff tags/release-1.0.1_2008-05-06_0815-EST..tags/release-1.0.1_2008-05-06_1200-EST | 09:10 |
fitzsim | basically, there should be a pretty easy way to map historical tag names to the git convention | 09:10 |
fitzsim | huh | 09:11 |
kanzure | the result of that diff: http://diyhpl.us/~bryan/irc/nanoengineer/random_diff.txt | 09:11 |
fitzsim | oh, I thought you meant they were no differences | 09:11 |
kanzure | frankly it seems like some minor cleanup | 09:12 |
kanzure | plus release notes.. that sort of thing | 09:12 |
fitzsim | anyway, I like the last comment on that Stack Overflow thread | 09:13 |
kanzure | i'm not familiar with git describe | 09:14 |
fitzsim | although, I just realized that doesn't really work for tags, only for maintenance branches | 09:14 |
fitzsim | s/tags/release tags/ | 09:14 |
kanzure | anywho i've pushed the tags i've created so far to nanoengineer-fixed.git if you want to test how long cloning that takes | 09:15 |
kanzure | huh cgit isn't showing any non-remote branches | 09:15 |
kanzure | ok i did git push --mirror /srv/git/nanoengineer-fixed.git so you should be able to clone now and get goodies back | 09:17 |
kanzure | or pull | 09:17 |
fitzsim | ok, I just re-cloned two minutes ago and didn't get any new branches | 09:18 |
fitzsim | shall I try again now? | 09:18 |
fitzsim | re-cloning... | 09:18 |
kanzure | yes | 09:18 |
kanzure | git push --mirror told me it created about 10 new branches on nanoengineer-fixed | 09:19 |
kanzure | however, i only see one branch (master) and three tags (three tags is right) http://diyhpl.us/cgit/nanoengineer-fixed/refs/?h=trunk | 09:19 |
fitzsim | yup, looks like that took | 09:19 |
fitzsim | nope | 09:20 |
kanzure | hm? | 09:20 |
fitzsim | $ git branch -a | 09:20 |
fitzsim | * master | 09:20 |
fitzsim | remotes/origin/HEAD -> origin/master | 09:20 |
fitzsim | remotes/origin/master | 09:20 |
kanzure | what abuot tags | 09:20 |
kanzure | about | 09:20 |
fitzsim | $ git tag -l | 09:22 |
fitzsim | Distribution | 09:22 |
fitzsim | cad | 09:22 |
fitzsim | sim | 09:22 |
fitzsim | 09:22 | |
-!- Redeemer [~lorddeeme@c-75-72-218-226.hsd1.mn.comcast.net] has quit [Ping timeout: 255 seconds] | 09:24 | |
kanzure | fitzsim: woops. looks like i need to checkout the remote branches | 09:26 |
kanzure | git checkout --track -b wware_jni_nd1 remotes/wware_jni_nd1 | 09:26 |
kanzure | and then when i'm sure i'm done delete all of the remote branches | 09:26 |
fitzsim | what does --track do? | 09:27 |
kanzure | well it's primarily to keep remote refs and local refs merged. i guess it's not needed here. | 09:30 |
fitzsim | the behavior of leaving out -b seems interesting | 09:32 |
fitzsim | I'm trying to understand why there are cad/ and sim/ tags and branches | 09:33 |
kanzure | presumably they wanted to keep cad/ and sim/ separated | 09:34 |
kanzure | but if you check any of the publicly available releases cad/ and sim/ are released together | 09:34 |
kanzure | btw you should remind me later to merge 'trunk' and 'master' | 09:36 |
fitzsim | ok | 09:39 |
kanzure | so it looks like for the tags that i converted earlier it's working on cgit? http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 09:39 |
fitzsim | yeah, I saw them in my cloned repo too | 09:40 |
kanzure | i'm wondering since i never did checkout those remote tags | 09:40 |
kanzure | whereas i'm having to checkout these other remote branches | 09:40 |
kanzure | ok. | 09:40 |
fitzsim | no, checkouts shouldn't be involved, come-to-think of it | 09:41 |
fitzsim | since they only affect the working directory | 09:41 |
kanzure | git checkout -b Distribution remotes/Distribution | 09:41 |
kanzure | then i get a local branch Distribution which i can push to nanoengineer-fixed.git | 09:41 |
kanzure | try pulling nanoengineer-fixed.git now | 09:42 |
kanzure | also the branches are now listed here: http://diyhpl.us/cgit/nanoengineer-fixed/refs/?id=bb38b9d350dfa97cf9f8a0f8d736aea5783d839e | 09:43 |
kanzure | er ignore the id= variable in that url. | 09:43 |
fitzsim | why doesn't http://pauldowman.com/2008/07/26/how-to-convert-from-subversion-to-git/ mention checkout? | 09:43 |
fitzsim | I think all SVN branches and tags will appear under tags/ | 09:44 |
kanzure | he was primarily dealing with tags | 09:44 |
kanzure | i don't think he had remote branches to deal with | 09:44 |
kanzure | other than the fact that his svn tags were stuffed into git branches | 09:45 |
kanzure | fitzsim: one of the comments points that out: "@Bojan Your one-liner assumes that all the branches have the form "tag/XXX" but that's not always true (e.g., when the SVN repo being converted already has other branches). I changed your one-liner to:..." | 09:46 |
fitzsim | ok | 09:47 |
fitzsim | does that one-liner handle what-were-SVN-branches too? | 09:48 |
kanzure | no that's just the convert-branches-that-were-svn-tags-into-git-tags | 09:48 |
kanzure | anyway, all of the branches are done on my end | 09:48 |
kanzure | now back to figuring out the tag naming scheme | 09:49 |
fitzsim | hmm, I'm getting confused by the conflicting terminologies | 09:49 |
kanzure | and whether or no to keep both release-1.0.1_2008-05-06_0815-EST and release-1.0.1_2008-05-06_1200-EST | 09:49 |
kanzure | hm? | 09:49 |
fitzsim | keep them both | 09:49 |
kanzure | why | 09:49 |
kanzure | they made a release and then released again a few hours later? | 09:49 |
kanzure | here's the diff: http://diyhpl.us/~bryan/irc/nanoengineer/random_diff.txt | 09:50 |
fitzsim | yeah, saw that | 09:50 |
fitzsim | I just think it's better to keep everything for historical reasons | 09:50 |
fitzsim | it seems arbitrary to fix that one | 09:50 |
kanzure | heh i'm just starting at the top of the list | 09:50 |
fitzsim | right, I understand | 09:51 |
fitzsim | I need to think about this some more though | 09:51 |
fitzsim | in particular I need to get clear on the different concepts | 09:51 |
kanzure | which concepts | 09:51 |
fitzsim | in SVN, they have: tags/ and branches/ | 09:51 |
-!- augur [~augur@74.174.236.120] has quit [Remote host closed the connection] | 09:51 | |
kanzure | tags are like a bookmark to a commit | 09:51 |
fitzsim | right | 09:51 |
fitzsim | say we have tags/foo and branches/bar | 09:52 |
fitzsim | in the SVN repo | 09:52 |
fitzsim | if you run git-svn to import that SVN repo, what's in the resulting git repo? | 09:52 |
fitzsim | git tags -l | 09:52 |
fitzsim | and git branch -a | 09:52 |
fitzsim | what will they report? | 09:53 |
kanzure | nothing | 09:53 |
kanzure | oops | 09:53 |
kanzure | git tags -l will report nothing | 09:53 |
kanzure | git branch -a will report http://diyhpl.us/~bryan/irc/nanoengineer/git-branch-a-git-svn.txt | 09:53 |
fitzsim | so in my example, it will report: | 09:55 |
fitzsim | remotes/bar | 09:55 |
fitzsim | remotes/tags/foo | 09:55 |
kanzure | yep | 09:56 |
fitzsim | now what does the "remotes" mean? | 09:57 |
kanzure | basically anything in remotes/ is a reference to the svn repository | 09:57 |
kanzure | oh wait i shouldn't say that actually | 09:58 |
kanzure | this might help you http://www.gitready.com/beginner/2009/03/09/remote-tracking-branches.html | 10:00 |
kanzure | i was going to say that they are an entry on git-remote but they are not (apparently) | 10:00 |
kanzure | i expected git-svn to setup everything so that it's just a complex git-remote-add call, but they are doing something different *shrug* | 10:01 |
fitzsim | can you paste the results of "git remote show remotes/cad" and "git remote show remotes/tags/cad" | 10:01 |
kanzure | from the vanilla-git-svn repo? | 10:01 |
fitzsim | yeah | 10:02 |
kanzure | "git remote" doesn't print anything so basically both of those will fail like "fatal: 'remotes/cad' does not appear to be a git repository" | 10:02 |
fitzsim | why is "git branch -a" listing them with the remotes/ prefix? | 10:04 |
kanzure | it's a "remotely tracked branch" | 10:05 |
fitzsim | does "git remote" print nothing, or an error? | 10:08 |
kanzure | nothing | 10:10 |
kanzure | git remote show remotes/tags/cad prints an error because "remotes/tags/cad" is not a remote ref.. uh | 10:10 |
kanzure | git remote add github kanzure@github.com:~/nanoenginer.git | 10:11 |
kanzure | git push github master | 10:11 |
kanzure | "github" is a remote ref in that example | 10:11 |
kanzure | for many of my repos i have a remote named 'diyhplus' for instance | 10:11 |
fitzsim | ok, so the "remotes/"-prefixed items in the output of "git branch -a" are not git remotes, they're git branches named with the "remotes/" prefix? | 10:13 |
kanzure | yes specifically they are the second type of git branches called remote branches or remotely-tracked | 10:13 |
kanzure | remotely-tracked branches, i mean. | 10:13 |
fitzsim | http://www.gitready.com/beginner/2009/03/09/remote-tracking-branches.html | 10:15 |
-!- klafka [~klafka@cpe-74-74-152-155.rochester.res.rr.com] has joined #hplusroadmap | 10:16 | |
kanzure | fitzsim: dunno if you saw but i linked you to that 10min ago | 10:16 |
fitzsim | oh, missed it | 10:18 |
fitzsim | why are SVN tags represented by git-svn as git tracking branches? | 10:19 |
fitzsim | https://github.com/nirvdrum/svn2git#readme | 10:26 |
fitzsim | why does "git branch -a" in their example there not show the "remotes/" prefixes? | 10:26 |
fitzsim | yeah, newer versions of git started adding the "remotes/" prefix around git 1.6 | 10:29 |
kanzure | ok so in nanoengineer-fixed.git there is a branch called "cad" | 10:32 |
kanzure | which was originally a tag in svn | 10:32 |
kanzure | http://diyhpl.us/cgit/nanoengineer/tree/tags/cad | 10:32 |
kanzure | wait. uh | 10:33 |
kanzure | yeah. okay. in svn it was a tag. | 10:33 |
kanzure | i converted it to a branch because xyz. i must be forgetting something | 10:33 |
kanzure | anyway, all of these folders in this branch are pretty crazy. they look like they should be tags too? | 10:34 |
fitzsim | hate to say it, but I think we may want to use svn2git instead | 10:34 |
kanzure | oh hm.. that's right, there was a branch and a tag | 10:35 |
kanzure | but tag:cad doesn't seem to have anything in nanoengineer-fixed.git http://diyhpl.us/cgit/nanoengineer-fixed/tag/?h=cad | 10:35 |
kanzure | ah here we go http://diyhpl.us/cgit/nanoengineer-fixed/tag/?id=cad | 10:36 |
kanzure | so this is tag:cad http://diyhpl.us/cgit/nanoengineer-fixed/commit/?id=8ab5e67b77a2a5760d1244b7aa83e54988e30e09 | 10:36 |
kanzure | from 2006-09-23 | 10:36 |
fitzsim | Does SVN fully-qualifies tag names? | 10:37 |
fitzsim | s/qualifies/qualify/ | 10:37 |
kanzure | oof i should know what that means but i don't | 10:37 |
kanzure | i'm just trying to differentiate between the tag and branch 'cad' (both of which existed in the svn repo) | 10:37 |
fitzsim | right | 10:37 |
fitzsim | that's the confusing part | 10:37 |
kanzure | tag cad: http://diyhpl.us/cgit/nanoengineer-fixed/commit/?id=8ab5e67b77a2a5760d1244b7aa83e54988e30e09 | 10:38 |
kanzure | branch cad: http://diyhpl.us/cgit/nanoengineer-fixed/tree/?h=cad | 10:38 |
fitzsim | I think svn2git would do everything we want for us, except that there may be tags and branches that have the same name, and I don't know how it would handle that | 10:38 |
fitzsim | In SVN, is "cad" a tag name? | 10:39 |
kanzure | tag cad has lots of edits to week_060923 annd week_060923 appears in branch cad too: http://diyhpl.us/cgit/nanoengineer-fixed/tree/week_060923?h=cad | 10:39 |
kanzure | yes in svn there is a tag "cad" and a branch "cad" | 10:39 |
kanzure | i think some of this might be due to their poor use of the svn repo | 10:39 |
kanzure | for instance, it makes no sense that they have a tag "cad" that points to a two year old revision | 10:40 |
kanzure | meanwhile branch "cad" has all sorts of crap that all look more like tags :P | 10:40 |
fitzsim | sort of, but it's ultimately SVN's fault for mixing up the tag/branch namespaces with the project file hierarchy namespace | 10:40 |
fitzsim | doing so is a recipe for this type of confusion | 10:40 |
kanzure | what are we confused about at the moment? besides "why does the 'cad' branch have all this crap in it / why is there a cad/sim/Distribution branch and a trunk branch" | 10:42 |
fitzsim | basically, ask yourself: how does git2svn figure out how deep to go when deriving a git tag name from the SVN repository's tags directory? | 10:43 |
fitzsim | I'll do an example | 10:44 |
fitzsim | working with http://diyhpl.us/cgit/nanoengineer/tree/ | 10:44 |
fitzsim | there's trunk/cad/src | 10:45 |
kanzure | in branch master? yes | 10:45 |
fitzsim | no, just pretend we're looking at the SVN repo | 10:46 |
kanzure | ok. | 10:46 |
fitzsim | trunk/cad/src | 10:46 |
fitzsim | branches/cad/bruce_070405/src | 10:46 |
fitzsim | tags/cad/Release_0-9-1_20070701/src | 10:46 |
kanzure | oh you know what part of this might be due to a bad cvs->svn conversion.. they might have started off with using cvs | 10:47 |
fitzsim | ideally, from that svn2git would produce: | 10:48 |
fitzsim | $ git branch | 10:48 |
fitzsim | * master | 10:48 |
kanzure | ok so you are suggesting that tags/cad/Release_0-9-1_20070701 should be the name of a tag and the tag name isn't just "cad" | 10:48 |
fitzsim | cad/bruce_070405 | 10:49 |
fitzsim | $ git tag -l | 10:49 |
fitzsim | cad/Release_0-9-1_20070701 | 10:49 |
fitzsim | right | 10:49 |
fitzsim | there's still a possibility that an SVN branch and an SVN tag would "resolve" to the same name | 10:51 |
kanzure | i'm not sure bruce_070405 is really a branch | 10:52 |
fitzsim | it's just an example, I could have picked any of the names in the branches/ directory | 10:53 |
kanzure | i mean, it's certainly a copy of the code plus some sort of revision set | 10:53 |
kanzure | hmm | 10:53 |
kanzure | i mean to say that i don't think anything in svn:branches/cad/* is a separate branch | 10:53 |
kanzure | you know, svn branches really suck | 10:55 |
fitzsim | yup :-) | 10:55 |
kanzure | i agree tags/cad/Releasewhatever should be a git tag with a name like vwhatever, sure | 10:57 |
kanzure | for branches, i'm not sure- it's not like there's any "this branch was branched from revision xyz" info laying around | 10:57 |
fitzsim | darn, the password I had for https://polosims-svn.svn.cvsdude.com/polosims/trunk/ doesn't work any more | 10:58 |
kanzure | we could just throw all of the branches into a git branch called "old crappy svn branches" | 10:58 |
kanzure | i really doubt anyone is ever going to go through all of those wware_050505 folders and sort through them | 10:59 |
kanzure | but they should still exist somewhere | 10:59 |
fitzsim | can you compare the recursive listing of https://polosims-svn.svn.cvsdude.com/polosims/trunk/ with http://diyhpl.us/cgit/nanoengineer/tree/ ? | 10:59 |
fitzsim | http://diyhpl.us/cgit/nanoengineer/tree/trunk , I mean | 11:00 |
fitzsim | grr, s/trunk/branches/ | 11:00 |
kanzure | it's exactly the same | 11:00 |
fitzsim | ok, just wanted to make sure git-svn didn't miss anything | 11:01 |
fitzsim | in that case, yeah, these SVN "branches" are strange | 11:01 |
fitzsim | tags can include a subset of the files on trunk at a given time | 11:01 |
fitzsim | but does it make sense for branches to track a subset of trunk? | 11:01 |
fitzsim | I think this is a side-effect of SVN's implementation | 11:02 |
kanzure | tags just point at a commit really | 11:02 |
fitzsim | can you try checkout out e.g. the bruce_070405 branch from SVN? | 11:03 |
fitzsim | s/checkout/checking/ | 11:03 |
kanzure | https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/bruce_070405/ | 11:04 |
fitzsim | no, I mean doing an svn checkout | 11:04 |
kanzure | why | 11:04 |
kanzure | the files are the same. | 11:05 |
fitzsim | but it doesn't just check out src, right? | 11:05 |
kanzure | what? | 11:05 |
kanzure | ok so you're saying if i checkout branches/cad/bruce_070405 that it will check out other stuff? O_o | 11:06 |
fitzsim | first of all, how do you refer to that branch on the SVN command line? | 11:06 |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has joined #hplusroadmap | 11:06 | |
kanzure | haha i don't use svn that often/ever so that's a good question :) | 11:06 |
kanzure | although on https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/bruce_070405/src/ it says the revision is 6933 for GLPane.py | 11:06 |
kanzure | and the log for GLPane.py in nanoengineer.git shows this: http://diyhpl.us/cgit/nanoengineer/log/branches/cad/bruce_070405/src/GLPane.py | 11:07 |
kanzure | under https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/ the commit message for bruce_070405 is "Moved remotely" (the same as the commit message for GLPane.py on nanoengineer.git) | 11:07 |
kanzure | "Moved remotely" is r9410 | 11:08 |
kanzure | huh there's actually a log on that file for the svn repo | 11:08 |
fitzsim | I think you should go through the excercise of checking out a particular branch from SVN, and keep it around, so that you can compare that with what you get for the analogous git sequence | 11:09 |
fitzsim | looking at the exploded SVN repository structure could confuse us with implementation details | 11:10 |
kanzure | git checkout cad -> "warning: refname 'cad' is ambiguous." uh oh. did i break this already? | 11:13 |
fitzsim | ok, that answer's one question :) | 11:13 |
kanzure | m? | 11:13 |
kanzure | hm? | 11:13 |
fitzsim | i.e. how does git handle tags and branches with the same name | 11:13 |
kanzure | i thought git checkout should be checking out a branch :( | 11:14 |
fitzsim | or a tag | 11:14 |
kanzure | bluh | 11:15 |
fitzsim | try differentiating with refs/heads/ and refs/tags/ prefixes | 11:15 |
kanzure | (well, at the moment, it's definitely checking out the branch) i wanted to go see if bruce_070405 has commit history in nanoengineer-fixed.git | 11:16 |
kanzure | wait.. why isn't bruce_070405 in http://diyhpl.us/cgit/nanoengineer-fixed/tree/?h=cad | 11:17 |
fitzsim | I don't know | 11:19 |
kanzure | ok well it's still checking out git branch 'cad' on my shell so we'll see if bruce_070405 is in branch 'cad' here too or not | 11:20 |
fitzsim | did you try the SVN branch experiment? | 11:24 |
-!- phreedom [~quassel@109.254.17.41] has quit [Quit: No Ping reply in 180 seconds.] | 11:31 | |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 11:34 | |
Utopiah_ | archels: the talk was recorded and is being uploaded | 11:37 |
-!- Utopiah_ is now known as Utopiah | 11:37 | |
-!- infomorph [~infomorph@99.94.155.220] has joined #hplusroadmap | 11:39 | |
-!- phreedom [~quassel@109.254.17.41] has quit [Quit: No Ping reply in 180 seconds.] | 11:41 | |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 11:41 | |
kanzure | fitzsim: no not yet. i'm still trying to figure out where bruce_070405 went. | 11:49 |
fitzsim | that is weird | 11:51 |
fitzsim | but I'd say don't spend too much time on that | 11:51 |
fitzsim | I think you'll have to re-do the repository again anyway | 11:51 |
fitzsim | check this out: | 11:52 |
fitzsim | http://www.spinics.net/lists/git/msg79353.html | 11:52 |
fitzsim | I think we want "the long way" mentioned there, in combination with what svn2git does | 11:52 |
fitzsim | mixing the two will be a trick though, since the "long way" steps happen right in the middle of the svn2git conversion process | 11:52 |
kanzure | oh i screwed up when i tried to fix the remote branches | 12:04 |
niftyzero | kanzure - let me know if you have git questions | 12:08 |
fitzsim | I think the sequence should be: | 12:08 |
fitzsim | git init | 12:08 |
fitzsim | add svn-remote section in .git/config as per http://www.spinics.net/lists/git/msg79353.html | 12:10 |
fitzsim | write a Ruby script to call fix_tags, fix_branches, fix_trunk and optimize_repos in https://github.com/nirvdrum/svn2git/blob/master/lib/svn2git/migration.rb | 12:11 |
fitzsim | IOW, do the same as the run! function there, but replace the clone step with our own custom step to handle the sub-directory-laden branches/tags layout of the NE-1 SVN repo | 12:12 |
niftyzero | you might just skip fixing the tags / branches, and even delete them if you don't care about painful historical details and just want to start new development from the tip of the trunk | 12:15 |
kanzure | fitzsim: actually i'm p. sure the main problem is that i corrupted the tag/branch namespace in nanoengineer-fixed.git | 12:16 |
fitzsim | niftyzero: we tried skipping fixing them, but the result is a very large and slow repository: http://diyhpl.us/cgit/nanoengineer | 12:16 |
fitzsim | s/large and// | 12:17 |
kanzure | fitzsim: if this is true then when i checkout -b blah remotes/cad i should get better history in nanoengineer-fixed2.git than i see here: http://diyhpl.us/cgit/nanoengineer/log/branches/cad/bruce_070405/src/GLPane.py | 12:17 |
kanzure | :( but i only get a handful of "Moved remotely" commits on refs/remotes/cad/bruce_070405/src/GLPane.py | 12:20 |
-!- fitzsim` [~fitzsim@bas3-toronto06-2925098524.dsl.bell.ca] has joined #hplusroadmap | 12:20 | |
kanzure | there are at least 30 revisions here: https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/bruce_070405/src/GLPane.py?view=log | 12:20 |
kanzure | hmm: r6872 "This commit was manufactured by cvs2svn to create branch 'bruce_070405'." | 12:21 |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 245 seconds] | 12:22 | |
-!- superkuh_ [hukrepus@unaffiliated/superkuh] has quit [Ping timeout: 245 seconds] | 12:22 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has quit [Ping timeout: 246 seconds] | 12:22 | |
niftyzero | you could kill all the branches and tags you don't care about, then garbage collect | 12:22 |
niftyzero | downloading... | 12:22 |
niftyzero | it looks like almost 100MB | 12:22 |
-!- superkuh_ [hukrepus@unaffiliated/superkuh] has joined #hplusroadmap | 12:22 | |
kanzure | niftyzero: 300MB for nanoengineer.git | 12:23 |
-!- fitzsim [~fitzsim@bas3-toronto06-2925098524.dsl.bell.ca] has quit [Ping timeout: 255 seconds] | 12:23 | |
-!- fitzsim` is now known as fitzsim | 12:23 | |
kanzure | fitzsim: http://gnusha.org/logs/2010-11-28.log | 12:23 |
kanzure | so i want to know why bruce_070405 files are not getting revision history under git | 12:24 |
kanzure | they have some, but it's mostly irrelevant stuff | 12:24 |
kanzure | and git log --follow just shows multiple "Moved remotely" commits | 12:25 |
fitzsim | <kanzure> fitzsim: if this is true then when i checkout -b blah remotes/cad i | 12:25 |
fitzsim | should get better history in nanoengineer-fixed2.git than i see | 12:25 |
fitzsim | here: | 12:25 |
fitzsim | http://diyhpl.us/cgit/nanoengineer/log/branches/cad/bruce_070405/src/GLPane.py | 12:25 |
fitzsim | <fitzsim> kanzure: but you're definitely going to have to deal specially with | 12:25 |
fitzsim | the tags/branches sub-directory layout | 12:25 |
fitzsim | <fitzsim> kanzure: there's no way around that | 12:25 |
fitzsim | <fitzsim> kanzure: git-svn by default expects tag names to be one level only | 12:25 |
fitzsim | under tags/ and branch names one level only under branches/ | 12:25 |
fitzsim | (that's what I pasted before the connection interruption) | 12:26 |
fitzsim | s/pasted/wrote/ | 12:26 |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has joined #hplusroadmap | 12:27 | |
kanzure | hmm | 12:27 |
kanzure | niftyzero: you shouldn't really bother with nanoengineer.git.. you can just browse the master branch's tree structure via cgit and get all the info you need | 12:33 |
kanzure | niftyzero: the idea is that i'm supposedly making a new .git that has everything fixed | 12:33 |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 12:34 | |
niftyzero | I'm just trying to give helpful git advice... ;) I already have a copy of the svn repo from before | 12:34 |
kanzure | you do? | 12:35 |
fitzsim | niftyzero: regarding omitting history in the conversion, I like kanzure's approach of a complete conversion because having all the history will be helpful in understanding the codebase | 12:35 |
niftyzero | ok | 12:35 |
kanzure | when did you get a copy of the svn repo | 12:36 |
fitzsim | niftyzero: do you have it in raw SVN form, or a checkout? | 12:36 |
fitzsim | and how recent is it? | 12:36 |
niftyzero | checkout | 12:36 |
fitzsim | oh, ok | 12:36 |
fitzsim | I have one too, with a guest account that doesn't work anymore | 12:37 |
fitzsim | but kanzure has a working account on the server that he's been running git-svn against | 12:37 |
niftyzero | cool | 12:37 |
fitzsim | a recent server-side backup of the repository would be useful for his experiments | 12:37 |
fitzsim | since it would save time/bandwidth | 12:38 |
fitzsim | kanzure: what do you think of the procedure I sketched out? | 12:38 |
kanzure | i'm looking around. | 12:39 |
-!- superkuh_ is now known as superkuh | 12:48 | |
kanzure | fitzsim: actually, i'm trying svn-all-fast-export | 12:49 |
fitzsim | kanzure: interesting. ok, let me know when you'd like me to test again | 12:53 |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 240 seconds] | 13:06 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 13:08 | |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has joined #hplusroadmap | 13:23 | |
-!- genehacker [~chatzilla@cpe-70-112-176-113.austin.res.rr.com] has joined #hplusroadmap | 13:28 | |
* kanzure twiddles his thumbs | 13:44 | |
kanzure | Utopiah: there's a french transhumanist association? | 13:45 |
kanzure | something about CA_AFT@yahoogroupes.fr | 13:45 |
Utopiah | yes | 13:45 |
Utopiah | Association Francaise Transhumaniste | 13:46 |
kanzure | oh i guess i mean http://fr.groups.yahoo.com/group/transhumanistes/ | 13:46 |
Utopiah | same | 13:46 |
Utopiah | AFAIK ~10members | 13:47 |
kanzure | not cool | 13:49 |
Utopiah | maybe it will change a lot after the 1st meeting as people will think the association can actually bring value | 13:52 |
kanzure | these two papers sound fun: | 13:56 |
kanzure | Neural substrates of envisioning the future | 13:56 |
kanzure | Episodic simulation of future events | 13:56 |
-!- bdesk_ [~argriffi@brcdesk18.statgen.ncsu.edu] has quit [Quit: leaving] | 14:01 | |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 245 seconds] | 14:18 | |
-!- augur [~augur@208.58.6.161] has joined #hplusroadmap | 14:21 | |
-!- genehacker [~chatzilla@cpe-70-112-176-113.austin.res.rr.com] has quit [Ping timeout: 265 seconds] | 14:23 | |
archels | http://telexlr8.wordpress.com/2010/11/22/suzanne-gildert-on-thinking-about-the-hardware-of-thinking-can-disruptive-technologies-help-us-achieve-uploading-teleplace-28th-november-2010-10am-pst/ | 14:40 |
archels | I see the post is revised, but there's no link to a video. | 14:40 |
-!- uniqanomaly [~ua@dynamic-78-8-83-20.ssp.dialog.net.pl] has quit [Read error: Operation timed out] | 14:41 | |
-!- uniqanomaly [~ua@dynamic-78-8-83-20.ssp.dialog.net.pl] has joined #hplusroadmap | 14:41 | |
-!- genehacker [~chatzilla@wireless-128-62-211-74.public.utexas.edu] has joined #hplusroadmap | 14:42 | |
Utopiah | archels: want a 1 line sum-up? | 14:42 |
archels | sure | 14:42 |
Utopiah | ASIC CPUs only might not be the right solution, there should be investment in alternatives and not just GPUs or FPGA but quantum computing and even other computing "solutions" | 14:43 |
Utopiah | (at least that was my understanding of the talk) | 14:43 |
archels | alright, that sounds a little disappointing | 14:47 |
kanzure | "They also made a range of social behavior errors (e.g., leaving a shop without paying, offering sexual favors in lieu of payment)." haha | 14:48 |
archels | Did they talk about memristors? | 14:48 |
Utopiah | briefly but mostly to complain about the recent simplistic IEEE spectrum article | 14:48 |
Utopiah | saying throwing memristors are the problem but without a proper design is idealistic | 14:49 |
Utopiah | s/are/at/ | 14:49 |
archels | ah, I'm glad they're not that short-sighted. :) | 14:50 |
-!- genehacker [~chatzilla@wireless-128-62-211-74.public.utexas.edu] has quit [Ping timeout: 276 seconds] | 14:53 | |
Utopiah | was about ~15people present | 14:54 |
archels | incredible | 14:54 |
Utopiah | AFAIK TeleXLR8 events are usually <20 participants so far | 14:56 |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has joined #hplusroadmap | 14:58 | |
-!- niftyzero1 [~niftyzero@gateway/tor-sasl/niftyzero1] has joined #hplusroadmap | 15:01 | |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has quit [Ping timeout: 264 seconds] | 15:04 | |
-!- augur [~augur@208.58.6.161] has quit [Remote host closed the connection] | 15:19 | |
-!- mheld [~mheld@pool-173-76-224-45.bstnma.fios.verizon.net] has quit [Quit: mheld] | 15:25 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 259 seconds] | 15:34 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 15:35 | |
-!- JaredW [~JaredW@122-59-91-107.jetstream.xtra.co.nz] has joined #hplusroadmap | 15:39 | |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has joined #hplusroadmap | 15:42 | |
-!- augur [~augur@129.2.129.33] has joined #hplusroadmap | 15:45 | |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has quit [Ping timeout: 264 seconds] | 16:06 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has quit [Read error: Connection reset by peer] | 16:08 | |
-!- opensanta_ [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has joined #hplusroadmap | 16:08 | |
-!- opensanta_ is now known as opensanta | 16:08 | |
-!- augur [~augur@129.2.129.33] has quit [Remote host closed the connection] | 16:19 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has quit [Ping timeout: 245 seconds] | 16:25 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 255 seconds] | 16:30 | |
kanzure | r6520 :( | 16:31 |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has joined #hplusroadmap | 16:31 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has quit [Remote host closed the connection] | 16:32 | |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has joined #hplusroadmap | 16:42 | |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Ping timeout: 245 seconds] | 16:47 | |
-!- augur [~augur@208.58.6.161] has joined #hplusroadmap | 16:59 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 17:07 | |
-!- elmom [~elmom@hoasnet-fe29dd00-137.dhcp.inet.fi] has quit [Ping timeout: 265 seconds] | 17:10 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 255 seconds] | 17:12 | |
-!- klafka [~klafka@cpe-74-74-152-155.rochester.res.rr.com] has quit [Quit: This computer has gone to sleep] | 17:14 | |
kanzure | r8050 | 17:14 |
-!- klafka [~klafka@cpe-74-74-152-155.rochester.res.rr.com] has joined #hplusroadmap | 17:14 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has joined #hplusroadmap | 17:34 | |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has joined #hplusroadmap | 17:37 | |
-!- mheld [~mheld@pool-173-76-224-45.bstnma.fios.verizon.net] has joined #hplusroadmap | 17:49 | |
-!- mheld [~mheld@pool-173-76-224-45.bstnma.fios.verizon.net] has left #hplusroadmap [] | 17:51 | |
kanzure | dbolser: bioinformatics.org used to have this huge wiki page on bioinfromatics covering a lot of topics.. do you know where it went? | 18:00 |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has joined #hplusroadmap | 18:02 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has quit [Read error: Connection reset by peer] | 18:02 | |
-!- opensanta [opensanta@c-76-110-86-5.hsd1.fl.comcast.net] has joined #hplusroadmap | 18:03 | |
kanzure | guess i wanted this: [edit] | 18:21 |
kanzure | bluh. http://www.bioinformatics.org/wiki/Bioinformatics_FAQ | 18:21 |
kanzure | back when it was all on one page | 18:21 |
-!- eridu [~eridu@gateway/tor-sasl/eridu] has quit [Quit: Leaving] | 18:43 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 18:49 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 250 seconds] | 18:53 | |
-!- wolfspraul [~wolfsprau@219-70-189-149.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 18:58 | |
-!- wolfspraul [~wolfsprau@219-70-189-149.cable.dynamic.giga.net.tw] has quit [Client Quit] | 18:59 | |
-!- wolfspraul [~wolfsprau@219-70-189-149.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 19:01 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 19:24 | |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 245 seconds] | 19:30 | |
-!- wolfspraul [~wolfsprau@219-70-189-149.cable.dynamic.giga.net.tw] has quit [Ping timeout: 276 seconds] | 19:36 | |
-!- genehacker [~chatzilla@cpe-70-112-176-113.austin.res.rr.com] has joined #hplusroadmap | 19:47 | |
-!- wolfspraul [~wolfsprau@219-70-169-15.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 19:47 | |
kanzure | hi genehacker | 19:48 |
-!- wolfspraul [~wolfsprau@219-70-169-15.cable.dynamic.giga.net.tw] has quit [Ping timeout: 255 seconds] | 19:54 | |
-!- phreedom [~quassel@109.254.17.41] has quit [Quit: No Ping reply in 180 seconds.] | 19:58 | |
-!- wolfspraul [~wolfsprau@219-70-186-177.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 20:10 | |
Redeemer | Kanzure, any news of anyone else looking to room up? | 20:20 |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 20:25 | |
-!- wolfspraul [~wolfsprau@219-70-186-177.cable.dynamic.giga.net.tw] has quit [Ping timeout: 255 seconds] | 20:26 | |
-!- phreedom [~quassel@109.254.17.41] has quit [Client Quit] | 20:29 | |
-!- codeshepherd [~Deepan@122.167.76.104] has joined #hplusroadmap | 20:31 | |
kanzure | Redeemer: nope :( | 20:33 |
-!- klafka [~klafka@cpe-74-74-152-155.rochester.res.rr.com] has quit [Quit: This computer has gone to sleep] | 20:33 | |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 20:35 | |
-!- wolfspraul [~wolfsprau@219-70-191-58.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 20:35 | |
fitzsim | kanzure: re: r8050 you're trying with svn-all-fast-export? | 20:40 |
-!- phreedom [~quassel@109.254.17.41] has quit [Ping timeout: 240 seconds] | 20:40 | |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 20:44 | |
-!- phreedom [~quassel@109.254.17.41] has quit [Quit: No Ping reply in 180 seconds.] | 20:50 | |
kanzure | fitzsim: yes | 20:51 |
kanzure | fitzsim: also it makes a full copy of the svn repo so i don't have to do this again | 20:51 |
kanzure | Exporting revision 9387 /branches/cad/avendor was copied from /trunk/cad/branches/avendor rev 9386 | 20:51 |
kanzure | "cad-avendor" is not a known branch in repository "polosims.git" | 20:51 |
kanzure | +++++++++++++++.Branch "cad-avendor" in repository "polosims.git" doesn't exist at revision 9387 -- did you resume from the wrong revision? 1 modifications from SVN /trunk/ to polosims.git/master 1125 modifications from SVN /branches/cad/avendor to polosims.git/cad-avendor done | 20:52 |
kanzure | ++Failed to write to process: Error writing to process | 20:52 |
kanzure | Aborted | 20:52 |
kanzure | real 419m59.592s | 20:52 |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 20:53 | |
-!- phreedom [~quassel@109.254.17.41] has quit [Client Quit] | 20:53 | |
fitzsim | kanzure: does it handle sub-directories in tags and branches? | 20:54 |
-!- wolfspraul [~wolfsprau@219-70-191-58.cable.dynamic.giga.net.tw] has quit [Ping timeout: 255 seconds] | 20:54 | |
-!- Redeemer [~lorddeeme@173-11-40-10-Minnesota.hfc.comcastbusiness.net] has quit [Ping timeout: 276 seconds] | 20:56 | |
-!- phreedom [~quassel@109.254.17.41] has joined #hplusroadmap | 21:00 | |
kanzure | fitzsim: yes it's supposed to | 21:04 |
fitzsim | yeah, from that failure message it looks like it names the cad/avendor branch "cad-avendor" which is good | 21:08 |
kanzure | that's from a custom rules file | 21:12 |
fitzsim | ok | 21:14 |
kanzure | http://diyhpl.us/~bryan/irc/nanoengineer/polosims.rules | 21:15 |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has joined #hplusroadmap | 21:16 | |
fitzsim | kanzure: for the branches/wware_jni_nd1/ rule I don't think you want the second path element | 21:19 |
kanzure | well, it's easily fiable | 21:20 |
kanzure | *fixable | 21:20 |
kanzure | both by rerunning the rules on the svn repo | 21:21 |
kanzure | or by later merging those wware_jni_nd1 branches together | 21:21 |
-!- flamoto [~root@bas5-barrie18-2925278090.dsl.bell.ca] has quit [Ping timeout: 272 seconds] | 21:21 | |
fitzsim | kanzure: in fact, wware_jni_nd1 looks like it is a standard one-level-deep branch name | 21:22 |
fitzsim | kanzure: is Distribution a directory or always just a tag? | 21:22 |
kanzure | Distribution is both a tag and a branch | 21:23 |
kanzure | in the global namespace | 21:23 |
kanzure | it also seems to be present quite often in various tags | 21:23 |
fitzsim | kanzure: I mean, would someone have created a top-level Distribution directory in their SVN working directory, then committed it? | 21:23 |
kanzure | dunno. | 21:23 |
kanzure | i should probably ask mark | 21:23 |
fitzsim | kanzure: I can't match the contents of wware_jni_nd1/Distribution with anything in trunk | 21:24 |
fitzsim | s%wware%branches/wware% | 21:24 |
kanzure | "/branches/wware_jni_nd1/" is a copy-with-history, auto-recursing | 21:26 |
kanzure | "/branches/wware_jni_nd1/Distribution/" rev 10115 did not match any rules; cannot continue | 21:26 |
fitzsim | yeah, I don't understand Distribution yet | 21:28 |
fitzsim | for example: root/branches/Distribution/wware_abmacosx_20060130/autoBuild.py | 21:29 |
fitzsim | is that "autoBuild.py in the top-level directory, on the Distribution-wware_abmacosx_20060130 branch"? | 21:29 |
kanzure | haha what? there's also weird things with names that look like branches under the branch "Distribution" ? | 21:30 |
fitzsim | yeah | 21:30 |
kanzure | fuck this | 21:31 |
kanzure | hm | 21:31 |
fitzsim | root/branches/wware_jni_nd1/Distribution looks like "a Distribution directory in the top-level directory, on the wware_jni_nd1 branch" | 21:31 |
QuantumG | chicken crimpys and coffee.. not as bad as I thought it would be | 21:32 |
kanzure | fitzsim: yeah, otherwise i would expect Distribution to have a more creative name | 21:32 |
kanzure | but that doesn't explain why there's a branch called Distribution with folders that look an awful lot like branches | 21:32 |
fitzsim | right | 21:34 |
fitzsim | was autoBuild.py an old file that was removed on the trunk at some point? | 21:35 |
kanzure | we could just ignore this wware_jni_nd1 thing | 21:35 |
fitzsim | I think the Distribution is just overloaded to mean "top-level directory" in some places, and "branch namespace prefix" in other places | 21:36 |
fitzsim | match /branches/wware_jni_nd1/ | 21:37 |
fitzsim | repository polosims.git | 21:37 |
fitzsim | branch wware_jni_nd1 | 21:37 |
fitzsim | end match | 21:37 |
kanzure | heh | 21:37 |
kanzure | yeah i'll try that | 21:38 |
fitzsim | and I think your second rule is good, just checking... | 21:38 |
kanzure | running the rules again.. (have to start from scratch each time) | 21:39 |
fitzsim | ninad_20070924_GGBC_experimental, ninad_20070924_GGBC_experimental_2 and ninad_20071211_selectMode_split look problematic | 21:41 |
kanzure | can you point me to those on http://diyhpl.us/cgit/nanoengineer/tree | 21:41 |
fitzsim | http://diyhpl.us/cgit/nanoengineer/tree/branches/cad/ninad_20070924_GGBC_experimental | 21:41 |
kanzure | hm | 21:42 |
kanzure | the rules will rename that to cad-ninad_20070924_GGBC_experimental_2 | 21:44 |
fitzsim | right, but it looks like it's actually a branch of a deeper sub-directory, whereas all the other branches (so far, still checking) contain a subset of the root directory | 21:44 |
fitzsim | branches/sim/* look fine too | 21:51 |
fitzsim | kanzure: does svn-all-fast-export know that to tranform e.g. "branches/cad/bruce_070405/src" to "cad/src on the cad-bruce_070405 branch" in the git repository? | 21:53 |
fitzsim | kanzure: in other words, does it understand that the src in "branches/cad/bruce_070405/src" is actually "cad/src", not "just src in the top-level directory"? | 21:54 |
kanzure | huh | 21:55 |
kanzure | uh | 21:55 |
kanzure | no i think not | 21:55 |
kanzure | but i'm not sure if that's the case though? | 21:55 |
kanzure | i mean, it makes sense, but i don't know if that's important | 21:56 |
fitzsim | yeah, maybe you're right that it's not the case | 21:57 |
fitzsim | I keep forgetting I'm looking at SVN repo implementation details | 21:58 |
kanzure | http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 21:58 |
kanzure | so uh | 21:58 |
kanzure | there's a branch called "cad" there | 21:58 |
kanzure | and it's basically still huge: http://diyhpl.us/cgit/nanoengineer-fixed/log/?h=cad | 21:59 |
kanzure | oops i should have linked to http://diyhpl.us/cgit/nanoengineer-fixed/tree/?h=cad | 21:59 |
fitzsim | yeah, there shouldn't be a cad branch | 22:00 |
kanzure | hrm. | 22:00 |
kanzure | i think all of those should be separate branches right? | 22:00 |
fitzsim | is nanoengineer-fixed the new-new repo? | 22:00 |
kanzure | i.e. cad-wware_qt4_snapshot_20061229_3pm (the bottom one on that last link) should be a branch | 22:00 |
kanzure | yes | 22:00 |
kanzure | feel free to clone it | 22:01 |
fitzsim | yeah, that last link should be a branch, for example | 22:01 |
fitzsim | it should appear in the drop-down at the top right of the cgit page | 22:01 |
kanzure | these branches are awful :( | 22:01 |
kanzure | intermediate_brink_of_A6_050815_before_povray_movie_commits | 22:02 |
kanzure | week_061102_beforeMainWindowFileSplit_ninad_061024_mainWindow | 22:02 |
kanzure | jrayhawk: :( | 22:03 |
fitzsim | kanzure: ok, this is an improvement, because the branches are at least listed | 22:04 |
kanzure | oh wait | 22:05 |
kanzure | these were tags originally http://diyhpl.us/cgit/nanoengineer/tree/tags/cad/ | 22:05 |
kanzure | oh cad-intermediate_brink_of_A6_050815_before_povray_movie_commits is listed as a tag on http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 22:05 |
kanzure | so basically the "cad" branch can be deleted.. i think | 22:07 |
kanzure | i think they might have committed to the "cad" branch even though it was mostly tags: http://diyhpl.us/cgit/nanoengineer-fixed/log/?h=cad | 22:07 |
fitzsim | cad-intermediate_brink_of_A6_050815_before_povray_movie_commits doesn't show up as a top-level directory | 22:07 |
kanzure | what? | 22:10 |
kanzure | it's a tag | 22:10 |
fitzsim | s/directory/tag/ | 22:10 |
kanzure | on http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 22:11 |
fitzsim | shouldn't it show up in the drop-down box? | 22:11 |
kanzure | http://diyhpl.us/cgit/nanoengineer-fixed/tag/?id=cad-intermediate_brink_of_A6_050815_before_povray_movie_commits | 22:11 |
kanzure | drop-down box is branches | 22:11 |
kanzure | no | 22:11 |
fitzsim | ok | 22:11 |
fitzsim | yeah, I see it there in refs | 22:11 |
kanzure | there's something like 200~ tags | 22:12 |
fitzsim | yeah | 22:12 |
fitzsim | so this is definitely an improvement | 22:12 |
kanzure | oh woops | 22:13 |
fitzsim | in that the branch and tag names are OK | 22:13 |
-!- Redeemer [~lorddeeme@75-168-97-236.mpls.qwest.net] has joined #hplusroadmap | 22:13 | |
fitzsim | but the branches aren't right yet | 22:14 |
fitzsim | and I think that's because of more weirdness in the SVN repo | 22:14 |
kanzure | no i screwed up and pushed to nanoengineer-fixed.git without first cleaning it out | 22:15 |
kanzure | so Distribution shouldn't actually be there any more | 22:15 |
kanzure | one sec | 22:15 |
fitzsim | ah, great | 22:15 |
fitzsim | so that's one issue that will go away | 22:15 |
fitzsim | kanzure: how did you get a password for polosims? | 22:16 |
fitzsim | I could really use one so I could actually see ViewVC and play with SVN | 22:16 |
fitzsim | right now I'm struggling with why the branches only show a subset of files, e.g. just the src directory for the most part | 22:16 |
kanzure | ok one sec | 22:17 |
fitzsim | in ViewVC I would expect to see what would be the result of a checkout from that branch | 22:17 |
kanzure | fitzsim: see pm | 22:17 |
kanzure | https://polosims-svn.home.cvsdude.com/ajax#login | 22:17 |
kanzure | fitzsim: new-new-new repo | 22:20 |
kanzure | http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 22:20 |
kanzure | what's going on here? http://diyhpl.us/cgit/nanoengineer-fixed/log/?h=cad-MERI i do not like seeing "Moved remotely" in my commit messages.. | 22:22 |
kanzure | it suggests that i'm missing something important | 22:22 |
kanzure | actually most of the branches have the "Moved remotely" commit message http://diyhpl.us/cgit/nanoengineer-fixed/refs/ | 22:26 |
kanzure | jrayhawk's alter ego says it's an artifact of svn not keeping track of how a branch is created, so it's to be expected/unfortunate | 22:30 |
fitzsim | yeah, I'm trying to figure out the commit graph of the bruce_070405 branch in the original SVN repo | 22:30 |
fitzsim | "Moved remotely" denotes a merge | 22:32 |
fitzsim | from branch to HEAD | 22:32 |
fitzsim | hehe "moved remotely" is sort of non-sensical as a phrase | 22:33 |
fitzsim | why is this SVN repository repo formed like this, with these branch subsets? | 22:34 |
kanzure | marijuana | 22:34 |
fitzsim | hahaha | 22:34 |
QuantumG | by the sounds of it | 22:35 |
fitzsim | kanzure: for example, compare: http://svn.apache.org/viewvc/subversion/trunk/ to http://svn.apache.org/viewvc/subversion/branches/1.4.x/ | 22:35 |
fitzsim | they both have the same file sets | 22:35 |
fitzsim | branches/1.4.x/ doesn't just show the different files | 22:35 |
kanzure | please realize that apache.org knows how to use svn | 22:36 |
kanzure | r868379 hahah | 22:37 |
fitzsim | ok, so you think NE commiters were using the repo in a non-standard way | 22:38 |
kanzure | yes | 22:39 |
fitzsim | old-nanorex/bruce_070405/src | 22:39 |
QuantumG | On April 11, 1954, absolutely nothing happened. | 22:39 |
fitzsim | $ svn log GLPane.py |tail | 22:40 |
fitzsim | r1029 | huaicai | 2005-02-08 22:21:40 -0500 (Tue, 08 Feb 2005) | 2 lines | 22:40 |
fitzsim | 22:40 | |
fitzsim | Use OpenGL color logic xor operation to do rubber-band window drawing. | 22:40 |
fitzsim | 22:40 | |
fitzsim | ------------------------------------------------------------------------ | 22:40 |
fitzsim | r976 | huaicai | 2005-02-01 13:21:00 -0500 (Tue, 01 Feb 2005) | 2 lines | 22:40 |
fitzsim | 22:40 | |
fitzsim | Revert GLPane.py to 1.103 | 22:40 |
fitzsim | 22:40 | |
kanzure | yes i know | 22:40 |
fitzsim | ------------------------------------------------------------------------ | 22:40 |
fitzsim | $ svn log GLPane.py |head | 22:40 |
fitzsim | ------------------------------------------------------------------------ | 22:40 |
fitzsim | r9410 | polosims_svn | 2007-08-09 22:36:05 -0400 (Thu, 09 Aug 2007) | 1 line | 22:40 |
fitzsim | Moved remotely | 22:40 |
fitzsim | 22:40 | |
fitzsim | how would they have done that, using svn commands? | 22:41 |
kanzure | basically they branched it / moved the file | 22:41 |
kanzure | one sec | 22:41 |
kanzure | 22:43:29 omg/jblake: they probably just checked out the whole repo and renamed a bunch of branches with svn mv or similar | 22:43 |
kanzure | 22:43:53 omg/jblake: renaming a branch won't confuse svn but it won't survive the import into git | 22:44 |
kanzure | 22:44:32 omg/jblake: i'm guessing that "moved remotely" commit was just a whole bunch of branch renames | 22:44 |
fitzsim | sounds about right | 22:46 |
fitzsim | though I'm not sure about the svn mv part | 22:46 |
fitzsim | maybe to start a branch they svn copy'd some files over | 22:46 |
kanzure | unfortunately we can't be sure of that | 22:47 |
kanzure | it might be true | 22:47 |
kanzure | or they might have copied-then-edited then commited to the new branch | 22:47 |
fitzsim | svn copy trunk/cad/src/graphics/widgets/GLPane.py trunk/cad/src/graphics/behaviors/confirmation_corner.py trunk/cad/src/command_support/modes.py branches/bruce_070405/src | 22:48 |
kanzure | and thanks to nanotechnology... http://www.youtube.com/watch?v=LFoC-uxRqCg | 22:48 |
kanzure | fitzsim: btw alot of the developers were using windows and tortoisesvn | 22:48 |
kanzure | so there's another lawyer of possible screwups involved there | 22:48 |
fitzsim | ah, ok, good to know | 22:48 |
fitzsim | then do multiple commits in branches/bruce_070405/src | 22:49 |
fitzsim | then svn mv the files back to HEAD when don | 22:49 |
fitzsim | e | 22:49 |
kanzure | huh? | 22:49 |
fitzsim | just speculating what the workflow would have been | 22:53 |
-!- mayko [~mayko@69-36-214-157.dynamic.dsl.skybest.com] has quit [Read error: Connection reset by peer] | 22:53 | |
fitzsim | svn copy trunk/[files subset] branches/<something>/src; multiple svn commit's on branch; svn mv branches/<something>/src/[files subset] trunk/[files subset] | 22:54 |
QuantumG | http://www.youtube.com/watch?v=WZ-1lfammjk oh god, the bridge is so good | 22:54 |
-!- mayko [~mayko@69-36-214-157.dynamic.dsl.skybest.com] has joined #hplusroadmap | 22:54 | |
kanzure | QuantumG: yeah i didn't know glory liu did another song | 22:54 |
kanzure | i'm worried singinst.org will do one about ai | 22:55 |
QuantumG | hehe | 22:55 |
kanzure | "this week on sesame street eliezer yudkowsky guest stars with big bird" | 22:55 |
QuantumG | one day smacking the side of my computer is not going to quiet down the fan.. I don't know what I will do on that day. | 22:57 |
jrayhawk | white lithium grease | 22:58 |
fitzsim | kanzure: how much scripting are you prepared to do? I think reconstructing everything is doable but tedious | 22:58 |
kanzure | i don't think it's worth it | 23:00 |
kanzure | you could attempt to convince me | 23:00 |
fitzsim | well, it would definitely be nice to have the entire commit history well-organized in a git repo | 23:03 |
fitzsim | grr, I wish they had at least maintained the file hierarchy on the branches | 23:03 |
kanzure | yes but i'm not sure these branches are worth that? | 23:03 |
QuantumG | the only accurate measure of code quality: wtfs-per-second | 23:03 |
fitzsim | but they just flattened everything into src | 23:04 |
* mayko is away: theres a shyness found in reason | 23:04 | |
fitzsim | kanzure: having a complete history can be very useful in explaining why things as they are currently | 23:05 |
fitzsim | kanzure: it would be one thing if the old SVN repo were still accessible somewhere | 23:05 |
fitzsim | but since the git repo will be "all we have" it would be a shame to completely get rid of that history | 23:06 |
kanzure | show me one of these branches that wasn't already integrated into HEAD | 23:07 |
fitzsim | maybe they're all merged into HEAD | 23:07 |
fitzsim | but the point is you lose all the individual branch diffs | 23:08 |
fitzsim | have you talked to Mark about getting a backup of the SVN repo? | 23:08 |
kanzure | no not yet | 23:09 |
fitzsim | if we had that on diyhpl.us then we could use git for all new development and the SVN repo for historical reference | 23:09 |
fitzsim | then we'd only have to convert trunk for the git repo | 23:10 |
fitzsim | that might be a nice halfway solution | 23:10 |
kanzure | what do you mean we'd have to convert trunk? | 23:11 |
kanzure | trunk was merged into master | 23:12 |
kanzure | http://diyhpl.us/cgit/nanoengineer-fixed/log/ | 23:12 |
fitzsim | yeah, that's all I meant | 23:12 |
fitzsim | yeah, it doesn't look possible to extract information from the SVN repository about these branches | 23:14 |
fitzsim | try: diff -r975:976 | 23:15 |
fitzsim | svn diff -r975:976 | 23:15 |
fitzsim | that's the commit that moved GLPane.py into branches/bruce_070405/src | 23:16 |
fitzsim | so a conversion script would basically have to figure out these "branch points" then replay the commits into the git repository on that branch | 23:18 |
fitzsim | http://stackoverflow.com/questions/1857857/convert-from-remote-hosted-svn-to-git | 23:20 |
fitzsim | the last comment there suggests no one's figured out how to do this properly | 23:20 |
kanzure | anywho we can just use my svn backup now | 23:23 |
epitron | rawr legacy crap | 23:23 |
kanzure | if you want that stored somewhere | 23:23 |
kanzure | epitron: your mom is legacy | 23:23 |
epitron | leggy? | 23:23 |
kanzure | same thing | 23:24 |
epitron | leg, i see. | 23:24 |
fitzsim | kanzure: you mean your checkout? | 23:24 |
epitron | is this that thing you guys were trying to put into git a long time ago? | 23:24 |
epitron | the one that shouldn't be in git? | 23:24 |
fitzsim | kanzure: or do you have a backup of the server-side SVN repo? | 23:24 |
kanzure | fitzsim: no not a checkout | 23:24 |
kanzure | epitron: no that was reprap | 23:25 |
epitron | ah yes | 23:25 |
fitzsim | kanzure: how hard would it be to host an anonymous SVN repo on diyhpl.us somewhere? | 23:26 |
fitzsim | kanzure: wait, if you had a backup why were you pointing to the server for the git conversion? | 23:27 |
kanzure | svn-all-fast-export made a backup of the svn repo | 23:27 |
kanzure | i'm not so sure about hosting an svn repo.. how about i just make it a tarball | 23:28 |
fitzsim | sure, that works | 23:28 |
fitzsim | does the svn-all-fast-export backup have complete history? | 23:28 |
kanzure | yes | 23:28 |
kanzure | it's basically the same as if mark would have given me a raw server dump | 23:28 |
kanzure | not svn co | 23:29 |
kanzure | epitron: what are you doing in here anyway | 23:29 |
fitzsim | huh, I didn't know you could do that with an SVN server, I thought there would be info it would hold back | 23:29 |
kanzure | epitron: what's up? | 23:29 |
epitron | lurking? | 23:29 |
kanzure | epitron: lurk more | 23:29 |
fitzsim | kanzure: yeah, a tarball of that is fine then | 23:29 |
kanzure | or something :P | 23:29 |
epitron | ROGER! | 23:29 |
kanzure | fitzsim: okay. | 23:29 |
fitzsim | kanzure: anyone who cares can download it and check the history on their own machine | 23:29 |
kanzure | sounds good to me | 23:29 |
fitzsim | kanzure: and new development can happen on the trunk -> git conversion | 23:30 |
kanzure | frankly a lot of these branches can just be deleted | 23:30 |
fitzsim | (which will be a lot smaller too, another benefit) | 23:30 |
kanzure | but if they are going to be deleted i want those deletes to be in the git repo | 23:30 |
fitzsim | hmm | 23:30 |
kanzure | i am particularly interested in branches that have actual version strings | 23:31 |
kanzure | fitzsim: do you think there's anything else i should do before i replace nanoengineer.git? | 23:32 |
fitzsim | wait, what I was thinking is some option to svn-all-fast-export to just say "ignore branches and tags" | 23:33 |
kanzure | why | 23:33 |
fitzsim | because the branches and tags it creates from this weird SVN history are essentially meaningless | 23:33 |
kanzure | i'm ok with this being in the git history | 23:34 |
kanzure | even if in a few days we choose to delete some branches | 23:34 |
kanzure | i'd especially like to ask mark about them | 23:34 |
fitzsim | for example, in the original repo, you can follow what went on on the branches/bruce_070405 | 23:34 |
kanzure | well no | 23:34 |
fitzsim | but in the git repo conversion there's no commit history on that branch | 23:35 |
kanzure | only one thing went on there (creating a new branch) | 23:35 |
fitzsim | no | 23:35 |
kanzure | nothing was done after | 23:35 |
-!- nsh[cgi] [520c731b@gateway/web/freenode/ip.82.12.115.27] has joined #hplusroadmap | 23:35 | |
fitzsim | that's what shows up in the git repo, just one src directory with some files in it | 23:36 |
nsh[cgi] | scott aaronson: opinions? | 23:36 |
kanzure | look: https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/bruce_070405/src/GLPane.py?view=log | 23:36 |
fitzsim | but try doing an svn checkout of branches/bruce_070405, cd to it, then do an svn log | 23:36 |
fitzsim | and you see the commit history on that branch | 23:36 |
-!- wolfspraul [~wolfsprau@219-71-202-171.cable.dynamic.giga.net.tw] has joined #hplusroadmap | 23:36 | |
kanzure | nsh[cgi]: i heard of him sometime in 2009.. didn't write anything down about him though :( | 23:36 |
QuantumG | soo.. hot | 23:36 |
kanzure | fitzsim: yes i know but look closer | 23:37 |
kanzure | fitzsim: the latest commit is r9410 | 23:37 |
QuantumG | might have to turn on the AC | 23:37 |
nsh[cgi] | he's apparently proved interesting results on the strength of post-selected quantum systems in terms of complexity classes | 23:38 |
nsh[cgi] | on the other hand, he has given talks on the same platform at eliezer | 23:38 |
kanzure | fitzsim: the latest productive commit on branch bruce_070405 corresponds to 75ec8d5f8309a3be19fee192782dbaa1d89918ec | 23:38 |
kanzure | fitzsim: and *then* r9410 which is where they created the branch | 23:39 |
kanzure | so they never committed after creating the branch | 23:39 |
kanzure | meaning everything was done previously | 23:39 |
kanzure | in fact, the diff from before it was a branch and to where it was a branch is nil | 23:39 |
kanzure | https://polosims-svn.browse.cvsdude.com/polosims/branches/cad/bruce_070405/src/GLPane.py?r1=6933&r2=9410 | 23:39 |
kanzure | if there was more than 4 files in src/ we could have turned bruce_070405 into a tag and deleted the branch in git | 23:41 |
kanzure | but it seems that someone just checked out some files and decided to make a new branch because the svn workflow is amazingly fucked up | 23:41 |
fitzsim | kanzure: so in the SVN repo, is there any meaning to branches/bruce_070405/src ? | 23:44 |
fitzsim | kanzure: when I run svn log in there, it prints the entire repository's entire log history | 23:45 |
fitzsim | kanzure: how do I log just what happened on that branch? | 23:45 |
kanzure | jrayhawk: you might know that off the top of your head? | 23:45 |
kanzure | fitzsim: the meaning of bruce_070405 is something probably that only bruce knows.. or it's stuck in an email somewhere where he explains why it's important that bruce_070405 exists | 23:46 |
kanzure | btw there's a mailing list nanorex-all that was private to the developers | 23:46 |
kanzure | if i can get my hands on those archives we're golden for a lot of things | 23:46 |
kanzure | (i also need to deploy bugzilla next, or do something with that data) | 23:46 |
fitzsim | but bruce_070405 represents some line of development, that branched from mainline, did some commits, then merged back to mainline, right? | 23:47 |
fitzsim | in the trunk -> git conversion we'll see the merge patch as a commit | 23:48 |
fitzsim | and so the discussion is whether or not the intermediate commits that happened between the branch point and the merge commit can be thrown away | 23:49 |
jrayhawk | "Logs follow copy history by default. Use --stop-on-copy to disable this behavior, which can be useful for determining branchpoints." | 23:49 |
fitzsim | for bruce_070405/src, --stop-on-copy just prints one log message | 23:50 |
fitzsim | "Moved remotely" | 23:51 |
kanzure | fitzsim: nobody commited to bruce_070405 ever | 23:51 |
fitzsim | does that mean this is a branchpoint with no subsequent commits on it? | 23:51 |
jrayhawk | apparently | 23:51 |
fitzsim | ok, weird | 23:51 |
fitzsim | do people generally delete SVN branches after they've been merged? | 23:52 |
QuantumG | never | 23:52 |
fitzsim | or will we see branches with branchpoints and merge points? | 23:52 |
QuantumG | unless someone writes "merged from <branch>" in the commit message then you don't know there's been a merge | 23:53 |
QuantumG | you can guess from the rev numbers, that's all | 23:53 |
fitzsim | ugh | 23:53 |
QuantumG | but if they merge and then keep committing on the branch, too bad | 23:53 |
QuantumG | and people regularly do shit like merge a branch onto another branch and fix all the conflicts.. in one rev | 23:54 |
QuantumG | my personal favourite is that svn has no "uncommit" .. you can revert your changes and commit again, but that's not exactly the same thing | 23:55 |
fitzsim | kanzure: ok, so bruce_070405 is an example of what you're talking about, where deleting the "branch" is essentially harmless | 23:56 |
kanzure | well, "mostly harmless" | 23:56 |
QuantumG | so if you want to enforce a policy like: don't check in unless your code compiles, asshole.. then too bad, cause once they commit you can't get rid of that revision and maintain a pristine commit history. | 23:56 |
fitzsim | let me try to find a branch with actual development on it | 23:57 |
QuantumG | whereas if they're using git you can email them and say "hey asshole, your code doesn't compile, git uncommit" | 23:58 |
QuantumG | of course, push and pull screws that anyway | 23:58 |
kanzure | fitzsim: cad-ninad_20070924_GGBC_experimental, cad-ninad_20070924_GGBC_experimental_2, cad-ninad_20071211_selectMode_split | 23:58 |
QuantumG | so yeah, it's a consistent problem and just slapping people is the only solution | 23:59 |
--- Log closed Mon Nov 29 00:00:07 2010 |
Generated by irclog2html.py 2.15.0.dev0 by Marius Gedminas - find it at mg.pov.lt!