Bug#1042889: vm: autopkgtest fails against Emacs 29.1
Hello,
On Mon 14 Aug 2023 at 10:47am +01, Ian Jackson wrote:
> Sean Whitton writes ("Bug#1042889: vm: autopkgtest fails against Emacs 29.1"):
>> vm's autopkgtest fails with Emacs 29.1, which latter is now in sid.
>>
>> https://ci.debian.net/packages/v/vm/testing/amd64/
>
> The regression was indeed caused solely by upstream renaming the
> variable that disables byte compilation, from
> native-comp-deferred-compilation-deny-list
> to
> native-comp-jit-compilation-deny-list
>
> ISTM that making that change without also honouring the old setting
> was an obviously bad idea. Upstream set us up for this failure.
> Anyway, I have fixed this issue in vm, by handling both variables.
> I expect the new vm package to migrate quickly.
Usually there's an alias added when variable names are obsoleted.
Thank you for the fix.
--
Sean Whitton
signature.asc
Description: PGP signature
Bug#1042889: vm: autopkgtest fails against Emacs 29.1
Sean Whitton writes ("Bug#1042889: vm: autopkgtest fails against Emacs 29.1"):
> vm's autopkgtest fails with Emacs 29.1, which latter is now in sid.
>
> https://ci.debian.net/packages/v/vm/testing/amd64/
The regression was indeed caused solely by upstream renaming the
variable that disables byte compilation, from
native-comp-deferred-compilation-deny-list
to
native-comp-jit-compilation-deny-list
ISTM that making that change without also honouring the old setting
was an obviously bad idea. Upstream set us up for this failure.
Anyway, I have fixed this issue in vm, by handling both variables.
I expect the new vm package to migrate quickly.
I see that there are other packages with failing tests, preventing
emacs from migrating. I haven't looked at them, but some of them may
be similarly afflicted.
Thanks,
Ian.
--
Ian JacksonThese opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
Bug#1042889: vm: autopkgtest fails against Emacs 29.1
Hello,
On Wed 02 Aug 2023 at 02:49pm +01, Ian Jackson wrote:
> Sean Whitton writes ("Bug#1042889: vm: autopkgtest fails against Emacs 29.1"):
>> vm's autopkgtest fails with Emacs 29.1, which latter is now in sid.
>
> Hi, Sean, as you see we're looking into this.
> I have some questions for you as an Emacs expert:
>
> Is byte-compilation known to be sometimes broken? Is there a
> recommended approach to problems caused by byte-compilation ?
The bytecompiler tends to get fussier with each Emacs release.
Usually it's worth fixing the problems with the code it identifies.
It's unlikely to be actually broken in a stable release of Emacs.
> We recently did an update to vm in Debian stable, to work around a
> critical problem with Emacs 28 (#1039105). The autopkgtest which is
> now failing is new - I introduced it to detect future bugs, which it
> seems to have done.
>
> The previous bug was related to byte-compilation and we "fixed" it by
> turning off byte-compilation for at least some of vm's files (in what
> I feel was rather an ad-hoc way, albeit an effective one).
>
> Or to put it another way, is it possible that this is a bug in emacs
> 29.1 and if so what is the best workaround ?
Before disabling anything, it seems worth looking at the code and seeing
if you can figure out why there's a void-variable error being emitted at
all.
--
Sean Whitton
signature.asc
Description: PGP signature
Bug#1042889: vm: autopkgtest fails against Emacs 29.1
Sean Whitton writes ("Bug#1042889: vm: autopkgtest fails against Emacs 29.1"):
> vm's autopkgtest fails with Emacs 29.1, which latter is now in sid.
Hi, Sean, as you see we're looking into this.
I have some questions for you as an Emacs expert:
Is byte-compilation known to be sometimes broken? Is there a
recommended approach to problems caused by byte-compilation ?
We recently did an update to vm in Debian stable, to work around a
critical problem with Emacs 28 (#1039105). The autopkgtest which is
now failing is new - I introduced it to detect future bugs, which it
seems to have done.
The previous bug was related to byte-compilation and we "fixed" it by
turning off byte-compilation for at least some of vm's files (in what
I feel was rather an ad-hoc way, albeit an effective one).
Or to put it another way, is it possible that this is a bug in emacs
29.1 and if so what is the best workaround ?
Thanks,
Ian.
--
Ian JacksonThese opinions are my own.
Pronouns: they/he. If I emailed you from @fyvzl.net or @evade.org.uk,
that is a private address which bypasses my fierce spamfilter.
Bug#1042889: vm: autopkgtest fails against Emacs 29.1
Source: vm Version: 8.2.0b-10 Severity: serious Dear maintainer, vm's autopkgtest fails with Emacs 29.1, which latter is now in sid. https://ci.debian.net/packages/v/vm/testing/amd64/ Thanks. -- Sean Whitton signature.asc Description: PGP signature
