From: Chris Stewart <stewart.chris1234@gmail•com>
To: Christian Decker <decker.christian@gmail•com>
Cc: "Martin Habovštiak" <martin.habovstiak@gmail•com>,
"Bitcoin Development Mailing List" <bitcoindev@googlegroups.com>
Subject: Re: [bitcoindev] [Proposal] 64-bit arithmetic in Script
Date: Wed, 14 May 2025 03:23:00 -0500 [thread overview]
Message-ID: <CAGL6+mFF8noQxXWy8PXBT9vSBQv0Hx3jn7cFtQMkv4PcnhyGyw@mail.gmail.com> (raw)
In-Reply-To: <CALxbBHW8YD-F2N9PYcAii5wXfEAPratQ6i6ke-i59pdoFczSoA@mail.gmail.com>
[-- Attachment #1: Type: text/plain, Size: 8830 bytes --]
Hi Christian,
Thank you for the interest in this proposal!
I’d like to invite you, Rusty, or any other contributors to provide an
update to the list on the status of GSR. The most recent public writing I’m
aware of is Rusty’s blog post
<https://rusty.ozlabs.org/2024/01/19/the-great-opcode-restoration.html>,
which is now around 18 months old. Are there any newer materials — such as
additional posts, WIP BIPs, or code — that we could review or experiment
with? Even rough drafts would be helpful for prototyping and discussion.
I’m not opposed to the broader goals of GSR, but I do think it’s a bit
ambitious. That’s partly why I’ve focused my efforts on isolating what I
believe is the most requested feature: 64-bit precision to enable amount
locks.
>arbitrary sized integers
It would be helpful to see some concrete examples of opcodes that would
*require* arbitrary precision, but wouldn’t be achievable with 64-bit
arithmetic. Looking at the Elements project, there are a couple of examples
— ECMULSCALARVERIFY
<https://github.com/ElementsProject/elements/blob/7ed597f4a5f713e33aac04c87c1a9da5ecd7d52c/src/script/interpreter.cpp#L2193>
and TWEAKVERIFY
<https://github.com/ElementsProject/elements/blob/7ed597f4a5f713e33aac04c87c1a9da5ecd7d52c/src/script/interpreter.cpp#L2219>
— which operate on 256-bit stack arguments. Notably, these opcodes don’t
support composition with existing arithmetic opcodes like OP_ADD or OP_SUB;
they simply verify cryptographic conditions. I would argue they do not
actually require supporting more precision in Script as the stack arguments
aren't parsed into CScriptNum.
It could be useful to have a list of these potential opcodes that could be
enabled in a single place to give other protocol developers an idea of what
is enabled by arbitrary precision.
>maybe you could join that effort for your use-cases too?
Where can one go to join the effort?
-Chris
On Tue, May 13, 2025 at 6:45 AM Christian Decker <decker.christian@gmail•com>
wrote:
> Hi Chris,
>
> quite an interesting proposal, but much like Martin I wonder if we
> shouldn't go beyond. For comparison Rusty's GSR comprises arbitrary
> sized integers and their associated operations, with fine-grained
> accounting of operations based on the operands size, so scripts stay
> performant and manageable.
>
> That proposal is much further along in both design and analysis, so
> maybe you could join that effort for your use-cases too?
>
> Cheers,
> Christian
>
> On Tue, May 13, 2025 at 11:06 AM Chris Stewart
> <stewart.chris1234@gmail•com> wrote:
> >
> > Hi Martin
> >
> > Thanks for your interest in the topic.
> >
> > >It mentions upgrading existing opcodes, which is a hardfork, not soft
> fork, at least without using a different leaf version. But it also mentions
> OP_SUCCESSX which are different opcodes
> >
> > I view 64-bit arithmetic as a feature of a wider set of consensus
> changes. Here is what I think the most likely deployment story is
> >
> > >This proposal could be deployed in conjunction with any of the new
> opcode proposals in the Motivation section using Tapscript's OP_SUCCESSx
> semantics.[18]
> >
> > The opcodes mentioned in the motivation section are OP_{IN,OUT}_AMOUNT,
> OP_VAULT, OP_CHECKCONTRACTVERIFY, OP_CTV
> >
> > As an example, this proposal could (hypothetically) be deployed in
> conjunction with OP_CCV. OP_CCV would take advantage of the OP_SUCCESSx
> semantics allowing us to redefine existing opcode's (OP_ADD,OP_SUB, etc)
> semantics.
> >
> > There’s been some discussion around deploying OP_CTV via a NOP opcode
> instead of OP_SUCCESSx. I think that would be a poor choice, as it wouldn’t
> allow new Script features to be shipped in parallel with the new opcode.
> >
> > I found this StackExchange post helpful in thinking through various
> deployment strategies for new Tapscript-based consensus upgrades.
> >
> > >I'd also love to see analysis why stop at 64 bits and not go all the
> way to 256 which could be useful for cryptography.
> >
> > In my view, there’s still a lot of uncertainty about cryptographic
> features being added to Script. There's increasing discussion around
> quantum computing, which raises the question of how much numerical
> precision we may eventually need. I'm not opposed to extending precision
> further—but if we go beyond 64 bits, why stop at 256? With OP_SUCCESSx,
> arbitrary precision becomes a real option.
> >
> > I chose 64 bits because it covers what’s needed for amount locks. If
> someone strongly believes that 64 bits isn't enough, I’d welcome a
> competing BIP and would be happy to review and discuss it with the author.
> >
> > >Anyway, pushing amounts on the stack would be great. Though I'm
> surprised you're only proposing the sum, not individual outputs. Why?
> >
> > Good question—and slightly out of scope for this BIP. Script doesn’t
> support looping, so it’s not straightforward to iterate over and sum all
> elements unless the transaction structure (i.e., number of inputs or
> outputs) is known in advance.
> > You can measure the number of stack elements with OP_DEPTH, but there’s
> no way to write something like OP_ADD [num-stack-elements-from-OP_DEPTH] to
> sum them all. I’m definitely open to hearing other approaches, though.
> >
> > -Chris
> >
> >
> >
> > On Mon, May 12, 2025 at 2:32 PM Martin Habovštiak <
> martin.habovstiak@gmail•com> wrote:
> >>
> >> Hi,
> >>
> >> the proposal seems to be quite confused about how it's going to do
> that. It mentions upgrading existing opcodes, which is a hardfork, not soft
> fork, at least without using a different leaf version. But it also mentions
> OP_SUCCESSX which are different opcodes. I think it needs some analysis.
> (leaf version seems better intuitively)
> >>
> >> I'd also love to see analysis why stop at 64 bits and not go all the
> way to 256 which could be useful for cryptography.
> >>
> >> Anyway, pushing amounts on the stack would be great. Though I'm
> surprised you're only proposing the sum, not individual outputs. Why?
> >>
> >> Good luck!
> >>
> >> Martin
> >>
> >> Dňa po 12. 5. 2025, 14:21 Chris Stewart <stewart.chris1234@gmail•com>
> napísal(a):
> >>>
> >>> This soft fork proposal extends the range of numeric operands in
> Script from -2^31+1 to 2^31-1, to -2^63+1 to 2^63-1. It further expands the
> result range for arithmetic operations from -2^63 to 2^63-1, to -2^127 to
> 2^127- 1.
> >>>
> >>> All existing opcodes[1] that interpret stack elements as numbers are
> upgraded to support 64-bit parameters.
> >>>
> >>> The existing number encoding format[2] and arithmetic semantics[3]
> from the original Bitcoin implementation are preserved, while enhancing the
> supported precision.
> >>>
> >>>
> https://github.com/Christewart/bips/blob/2025-03-17-64bit-pt2/bip-XXXX.mediawiki
> >>>
> >>> The purpose for this BIP is to lay the groundwork for introducing
> amounts into Script. This document takes no opinion on how this is done.
> >>>
> >>> I've prototyped a few different proposals now introducing amount locks
> into Script[0][1] and feel like this proposal is stable enough for serious
> review.
> >>>
> >>> -Chris
> >>>
> >>> [0] -
> https://delvingbitcoin.org/t/op-inout-amount/549/4?u=chris_stewart_5
> >>>
> >>> [1] -
> https://delvingbitcoin.org/t/op-inout-amount/549/5?u=chris_stewart_5
> >>>
> >>>
> >>>
> >>> --
> >>> You received this message because you are subscribed to the Google
> Groups "Bitcoin Development Mailing List" group.
> >>> To unsubscribe from this group and stop receiving emails from it, send
> an email to bitcoindev+unsubscribe@googlegroups•com.
> >>> To view this discussion visit
> https://groups.google.com/d/msgid/bitcoindev/CAGL6%2BmH%2B9iq5_SR-Fa5zVZRoTpHasX7xoprYeJZRd5D80J1GqA%40mail.gmail.com
> .
> >
> > --
> > You received this message because you are subscribed to the Google
> Groups "Bitcoin Development Mailing List" group.
> > To unsubscribe from this group and stop receiving emails from it, send
> an email to bitcoindev+unsubscribe@googlegroups•com.
> > To view this discussion visit
> https://groups.google.com/d/msgid/bitcoindev/CAGL6%2BmHv%2Bkn6SU9pf0Rz3FmM5OfcsmEtqGBRJ7Upb-b0MofS5A%40mail.gmail.com
> .
>
--
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups•com.
To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/CAGL6%2BmFF8noQxXWy8PXBT9vSBQv0Hx3jn7cFtQMkv4PcnhyGyw%40mail.gmail.com.
[-- Attachment #2: Type: text/html, Size: 11300 bytes --]
next prev parent reply other threads:[~2025-05-14 9:36 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-05-12 16:15 Chris Stewart
2025-05-12 19:32 ` Martin Habovštiak
2025-05-13 9:03 ` Chris Stewart
2025-05-13 11:44 ` Christian Decker
2025-05-14 8:23 ` Chris Stewart [this message]
2025-05-14 8:27 ` Christian Decker
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAGL6+mFF8noQxXWy8PXBT9vSBQv0Hx3jn7cFtQMkv4PcnhyGyw@mail.gmail.com \
--to=stewart.chris1234@gmail$(echo .)com \
--cc=bitcoindev@googlegroups.com \
--cc=decker.christian@gmail$(echo .)com \
--cc=martin.habovstiak@gmail$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox