--- Log opened Wed Sep 22 00:00:14 2021
02:20 -!- siv2r093008922 [~siv2r@103.77.37.176] has quit [Quit: The Lounge - https://thelounge.chat]
03:23 -!- siv2r093008922 [~siv2r@103.77.37.176] has joined #secp256k1
05:15 -!- siv2r093008922 [~siv2r@103.77.37.176] has quit [Quit: The Lounge - https://thelounge.chat]
05:19 -!- siv2r093008922 [~siv2r@103.77.37.176] has joined #secp256k1
08:34 < siv2r[m]> Hello everyone,
08:34 < siv2r[m]> I am getting the following error on running `make` after `configure --enable-coverage`. Error: https://pastebin.ubuntu.com/p/64tFWjdsm3/
08:34 < siv2r[m]> But I don't get any error on `make` after `configure` (with no options). I am using ubuntu 20.04. 
08:36 < siv2r[m]> s/But I don't get any error on `make` after `configure` (with no options). I am using ubuntu 20.04./But I don't get any error on `make` after `configure` (with no options). I am using ubuntu 20.04./
08:36 < siv2r[m]>  * Hello everyone,
08:36 < siv2r[m]> I am getting the following error on running `make` after `configure --enable-coverage`. 
08:36 < siv2r[m]> Error: https://pastebin.ubuntu.com/p/64tFWjdsm3/
08:36 < siv2r[m]> But I don't get any error on `make` after `configure` (with no options). I am using ubuntu 20.04.
08:47 < nickler> siv2r[m]: oh, this is a bug in libsecp that was introduced in a recent commit. fe_verify can only be used if VERIFY is defined.
08:48 < nickler> The CI doesn't test the --enable-coverage flag. Afaik there's no reason why it shouldn't.
08:51 < nickler> Ah I remember that there's a WIP PR that makes CI produce coverage reports which would have caught this.
08:57 < siv2r[m]> Will it be helpful if I created an issue reagarding this?
08:59 < siv2r[m]> <nickler> "siv2r: oh, this is a bug in..." <- Also, can you send the link to this commit? I can try looking into this issue. I want to get started with contibuting to secp. I think this bug might be a good place to start.
09:13 < nickler> siv2r[m]: yes that would be helpful. This should get fixed.
09:14 < nickler> https://github.com/bitcoin-core/secp256k1/commit/20abd52c2e107e79391a19d2d2f8845e83858dea
09:17 < gmaxwell> those fe_verifies should probably be removed. (if someone really wants to worry that from-storage is returning invalid field elements, it could be added to from_storages _VERIFY code)
10:01 < siv2r[m]> <nickler> "https://github.com/bitcoin-core..." <- Thanks!
11:26 -!- siv2r093008922 [~siv2r@103.77.37.176] has quit [Quit: The Lounge - https://thelounge.chat]
12:21 -!- roconnor [~roconnor@host-45-58-209-171.dyn.295.ca] has joined #secp256k1
12:22 < roconnor> the call to fe_verify needs to be replaced with a check that the values are normalized.
12:24 < roconnor> The _from_storage simply asserts that the values are normalized, (i.e. it is a precondition that the values from_storage were built from normalized values).
12:24 < roconnor> But the point of this test to verify that these conditions are indeed true.
12:29 < gmaxwell> roconnor: thats why I suggested that the from_storage code might better be augmented with a test so that in a verify build every fromstorage load fe_verifies.
12:30 < roconnor> That makes sense.
12:31 < roconnor> BTW, while I totally respect it was a mistake to call fe_verify here, ... how is it possible to run test.c without being in VERIFY mode?
12:31 < roconnor> wouldn't that do a bad job in testing things?
12:32 < gmaxwell> roconnor: it's a product of a coverage build, which has to turn off verify to avoid falsely killing the coverage results with enormous numbers of unreachable verify failure branches.
12:32 < sipa> for coverage analysis you don't really care about it testing anything
12:33 < roconnor> I see.  Interesting.
12:33 < gmaxwell> it's only the special case of coverage where that happens. And also the tests should still be checking stuff, because they use CHECK macro which still runs. They just don't get the full verify instrumentation.
12:33 < roconnor> anyhow, sorry for breaking things.
12:33 < gmaxwell> Also, as an aside I think it's useful to run tests without the verify instrumentation because it may help hide compiler bugs.
12:33 < roconnor> As usual, I blame my reviewers. ;)
12:33 < gmaxwell> don't be, not your fault.
12:34 < gmaxwell> right.
16:43 -!- roconnor [~roconnor@host-45-58-209-171.dyn.295.ca] has quit [Ping timeout: 252 seconds]
21:56 -!- belcher [~belcher@user/belcher] has quit [Ping timeout: 252 seconds]
22:08 -!- belcher [~belcher@user/belcher] has joined #secp256k1
22:35 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 246 seconds]
22:38 -!- luke-jr [~luke-jr@user/luke-jr] has joined #secp256k1
--- Log closed Thu Sep 23 00:00:15 2021