Bug#1042889: vm breakage with Emacs 29
Dirk Eddelbuettel writes ("Bug#1042889: vm breakage with Emacs 29"): > Yes given the previous bug report and fix ensuring no byte-compilation takes > place for vm may be best. At least until someone (upstream? another dev?) > understand why it breaks vm. FTR, I investigated this a bit and I discovered that the regression is probably because native-comp-deferred-compilation-deny-list has been renamed to native-comp-jit-compilation-deny-list I think we should ahve a patch that works with both Emacs versions (since I don't want to complicate backports etc.); currently I have this: (eval-after-load "comp" '(dolist deny-list '('native-comp-deferred-compilation-deny-list 'native-comp-jit-compilation-deny-list) (if (boundp deny-list) (add-to-list deny-list "/vm.*\.el" but it is wrong in some way that my limited lisp-fu wasn't able to diagnose in the limited time I have right now. I will look at this again RSN. 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 breakage with Emacs 29
On 2 August 2023 at 14:41, Ian Jackson wrote: | Dirk Eddelbuettel writes ("Re: vm breakage with Emacs 29"): | > On 2 August 2023 at 13:17, Ian Jackson wrote: | > | Hi. Since you were helpful with #1039105 "Fails to start with Emacs | > | 28" I thought I would draw your attention to #1042889 | > | "vm: autopkgtest fails against Emacs 29.1" [0] | > | | > | I won't have time to look at this until next week, probably. Any help | > | or background research would be greatly appreciated. We need to fix | > | this to avoid vm getting autoremoved. | > | | > | I did have a quick look at the test log [1] and the failure looks | > | genuine. I suggest we do any further diagnosis in the bug. | > | > The band-aid I found and submitted for #1039105 (ie per Fedora's tracker, | > "just do not byte compile") seems apt here, no? I still do not really read | > (or, for that matter, write) elisp but it seems to complain about byte code. | > | > So I would try two things: | > - turn off elisp byte compilation as in #1039105 | | The patch from #1039105 is still in the package. Ok. | Do we need to add to a list of files in it, or something, do you That was my hunch. | think ? Maybe it would be best to disable byte compilation | completely. Yes given the previous bug report and fix ensuring no byte-compilation takes place for vm may be best. At least until someone (upstream? another dev?) understand why it breaks vm. Dirk | | One thing that would be useful would be for someone to try out emacs | and vm in a sid chroot; that would confirm that this isn't a spurious | test failure (or confirm that it is). | | 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. -- dirk.eddelbuettel.com | @eddelbuettel | e...@debian.org
Bug#1042889: vm breakage with Emacs 29
Dirk Eddelbuettel writes ("Re: vm breakage with Emacs 29"): > On 2 August 2023 at 13:17, Ian Jackson wrote: > | Hi. Since you were helpful with #1039105 "Fails to start with Emacs > | 28" I thought I would draw your attention to #1042889 > | "vm: autopkgtest fails against Emacs 29.1" [0] > | > | I won't have time to look at this until next week, probably. Any help > | or background research would be greatly appreciated. We need to fix > | this to avoid vm getting autoremoved. > | > | I did have a quick look at the test log [1] and the failure looks > | genuine. I suggest we do any further diagnosis in the bug. > > The band-aid I found and submitted for #1039105 (ie per Fedora's tracker, > "just do not byte compile") seems apt here, no? I still do not really read > (or, for that matter, write) elisp but it seems to complain about byte code. > > So I would try two things: > - turn off elisp byte compilation as in #1039105 The patch from #1039105 is still in the package. Do we need to add to a list of files in it, or something, do you think ? Maybe it would be best to disable byte compilation completely. One thing that would be useful would be for someone to try out emacs and vm in a sid chroot; that would confirm that this isn't a spurious test failure (or confirm that it is). 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.