Release tracking PR: secp256k1-sys 0.10.0

[tcharding]

Vendor latest tagged version of secp256k1 and prepare for release:

• Bump the version number to 0.10.0
• Run the vendor script (vendoring secp256k1 0.4.1)
• Update lock files
• Add changelog entry
• Depend on new version in secp256k1/Cargo.toml

[tcharding]

Nothing is ever as simple as it seems, if we do this we will have to do another release of secp-sys as well.

[apoelstra]

Heh, we might as well leave secp-sys at 1.48 then. It's okay if it's behind the other crates.

[tcharding]

Ok, I'll spilt out another CI job for 1.48 just for secp-sys

[tcharding]

Bother, cargo 1.48 tries to parse the top level manifest still so chokes on the edition = 2021 even though we are only testing secp256k1-sys.

[tcharding]

Just in case you lost context for this PR with the name changes, the problem with leaving secp256k1-sys on 1.48.0 was that cargo test didn't run in the secp256k1-sys/ directory because cargo attempts to parse to top level manifest.

[sanket1729]

One comment. Rest looks good

[apoelstra]

Nothing is ever as simple as it seems, if we do this we will have to do another release of secp-sys as well.

Actually I'm a bit confused -- what is "this" here? If you mean, revendoring libsecp requires us to release a new version of secp-sys, then yes, this is true. And we ought to do that before releasing a new version of rust-secp. And if we're doing a release, why wouldn't we bump the MSRV at the same time?

I had initially read your comment as "if we bump the MSRV then we need to release a new secp-sys version, which we could avoid if we didn't bump the MSRV". But it seems like we need a new version regardless, because of the revendoring (which we should definitely do, there have been some further sidechannel defenses added upstream).

[apoelstra]

Oh, I understand, we're revendoring only to change the version number.

Why don't we just update to the latest version of libsecp256k1?

[tcharding]

Why don't we just update to the latest version of libsecp256k1?

I wondered that myself, was just artificially limiting scope, and I was lazy, I see now that lates is 0.4.1 - we should definitely upgrade to that. Will have another go.

[apoelstra]

Sounds good -- I think we should update to 0.4.1, do a new secp-sys release, bump the MSRV of secp-sys while we're at it, and then we have a release that is clearly worthwhile and no questions about whether we should try to hack around the need for it.

[tcharding]

MSRV change is over in #693 now, I'll do the vendoring here still because its tied into the version number. And use 0.4.1

[tcharding]

Doing #693 separately may have been an overkill but now I've used it in the CHANGELOG, it serves some purpose (maybe) as a historical record.

[apoelstra]

When I run the script locally I find four files that you committed that aren't committed on my end:

        supprimé :        depend/secp256k1/TAGS
        supprimé :        depend/secp256k1/src/libsecp256k1-config.h
        supprimé :        depend/secp256k1/src/libsecp256k1-config.h.in
        supprimé :        depend/secp256k1/src/stamp-h1

(Supprimé means "deleted" in French.)

This happens both directly on my system and in my local CI.

[tcharding]

No worries, I'm not going to get to this today but I will at some stage.

You speak French as well as English with no accent eh, that explains your grammar skills.

[apoelstra]

I can read French, though I am missing a lot of vocabulary and struggle to understand even simple news articles. I keep my locale set to fr_CA mostly to fuzz my software and make sure it doesn't do something crazy/broken with a non-English local.

But I cannot speak or write French. :)

[tcharding]

I keep my locale set to fr_CA mostly to fuzz my software and make sure it doesn't do something crazy/broken with a non-English local.

Lol, things like this remind me you are a good bloke to have on the team.

[tcharding]

Is there a reason we haven't merged this PR and release secp256k1-sys v.010.0 - is this waiting on anything from me?

[tcharding]

Rebased on master so we can merge this PR then #685 goes right on top of it.

[apoelstra]

Is there a reason we haven't merged this PR and release secp256k1-sys v.010.0 - is this waiting on anything from me?

I was waiting for it to show up in my github notifications again, indicating that everything this was blocked on had merged, and that you'd noticed so I didn't have to :).

[apoelstra]

When I run the script locally I find four files that you committed that aren't committed on my end:

        supprimé :        depend/secp256k1/TAGS
        supprimé :        depend/secp256k1/src/libsecp256k1-config.h
        supprimé :        depend/secp256k1/src/libsecp256k1-config.h.in
        supprimé :        depend/secp256k1/src/stamp-h1

(Supprimé means "deleted" in French.)

This happens both directly on my system and in my local CI.

All of these files are still present.

[tcharding]

Jee wiz, spent so much time talking about your French skills I forgot to actually do the fix ... its hard to get good help.

[tcharding]

Should be right now man.

[apoelstra]

ACK 2bba8f9

[apoelstra]

Tagged and published.