Bug#1017063: fixed in gjs 1.73.1-2

2022-09-25 Thread Jeremy Bicha
On Sun, Sep 4, 2022 at 11:42 AM Paul Gevers  wrote:
> On Sat, 13 Aug 2022 01:48:54 + Debian FTP Masters
>  wrote:
> >  gjs (1.73.1-2) unstable; urgency=medium
> >  .
> >* Rebuild against latest mozjs91 (Closes: #1017063)
>
> You forgot to enforce this with a *versioned* build dependency and now
> armhf got build with the old version.
>
> But anyways. This smells a bit like an ABI breakage. Did something go
> wrong with mozjs91? And maybe we want to prevent issues during partial
> upgrades, then mozjs91 should break the old version of gjs.
>
> Normally no-change rebuilds are handled as binNMU's by the Release Team,
> is there any particular reason why you didn't do that?

Sorry, I didn't see your message until now.

I believe mozjs91 started failing to build on arm with gcc-12. Perhaps
that was related to this breakage or maybe Mozilla changed things a
bit more than usual in the 91.12 ESR release.

By not bumping the Build-Depends, gjs was able to build everywhere.
But then it couldn't migrate because on most architectures it got a
Depends on the new mozjs so that didn't really help.

Anyway, this is all fixed in Unstable as gjs now depends on mozjs102
and mozjs102 now builds on all release architectures.

I'll try to remember that a binnmu would have been better in this kind
of situation.

Thank you,
Jeremy Bicha



Bug#1017063: fixed in gjs 1.73.1-2

2022-09-03 Thread Paul Gevers

Hi Jeremy,

On Sat, 13 Aug 2022 01:48:54 + Debian FTP Masters 
 wrote:

 gjs (1.73.1-2) unstable; urgency=medium
 .
   * Rebuild against latest mozjs91 (Closes: #1017063)


You forgot to enforce this with a *versioned* build dependency and now 
armhf got build with the old version.


But anyways. This smells a bit like an ABI breakage. Did something go 
wrong with mozjs91? And maybe we want to prevent issues during partial 
upgrades, then mozjs91 should break the old version of gjs.


Normally no-change rebuilds are handled as binNMU's by the Release Team, 
is there any particular reason why you didn't do that?


Paul


OpenPGP_signature
Description: OpenPGP digital signature