[bug#67512] [PATCH v4 3/4] gnu: Add wasm packages.

[Ian Eure]

Clément Lassieur <clement@lassieur.org> writes:

> On Wed, Feb 21 2024, Liliana Marie Prikler wrote:
> > Am Dienstag, dem 20.02.2024 um 18:18 -0800 schrieb Ian Eure:
> > > Clément Lassieur <clement@lassieur.org> writes:
> > >
> > > > > Are you saying you want a process like:
> > > > > 
> > > > > 1a. Get wasm toolchain stuff merged.
> > > > > 1b. Get Librewolf merged without WASM sandboxing.
> > > > > 2. Update icecat, torbrowser, mullvad, and librewolf to 
> > > > > use 
> > > > > WASM sandboxing.
> > > > 
> > > > Excatly.  1b can be done after 1a, or before 1a.
> > > > 
> > >
> > > Is there a technical reason why landing WASM sandboxing 
> > > support 
> > > for all browsers in the same patch is desirable?  I can intuit 
> > > none, and as I’m disinclined to either roll back portions of 
> > > my 
> > > existing patchset, or work on other browsers, the proposal is 
> > > disagreeable.
> > I think this ordering is w.r.t. *patch sets*, not patches.  I 
> > wouldn't
> > suggest dropping four packages into one patch.
>
> Indeed I've never said it should be done in one patch.  I said 
> one-shot
> as in ‘symmetrical’: the work required to add Wasm to our 
> browsers
> should be more or less the same for all browsers, and code 
> duplication
> should be avoided.

Forgive me for my imprecision, and thank you for the explanation. 
Unfortunately, the distinction makes little difference to me, as 
it still would require me to do work I’m unwilling to do.  My 
unwillingness has less to do with the amount of work than its 
scope: My goal is to get LibreWolf into Guix, and I simply have no 
desire or motivation to work on other browsers.

I think the best course of action is to reduce scope by removing 
the WASM component of this patch series entirely.  I’d send a new 
patch series without the WASM toolchain packages, and with WASM 
sandboxing disabled in the LibreWolf package.  The official 
LibreWolf binaries don’t appear to have this enabled, so no 
hardening would be sacrified vs. LibreWolf installed any other 
way.  And since I’m not the original author of the WASM packages, 
and not well-positioned to address problems with them, omitting 
them seems likely to circumvent difficulties in the review process 
and support of those.

What do you think?

Thanks,

 — Ian