Control: tags 1138991 confirmed upstream

Rene Engelhard wrote...

> While I still tihnk it shouzld take .zip as what it is, file is wrong
> here. And since I didn't see a bug filed by the strip-determinism people
> I am doing that right now. I at least didn't see one, forgive me if I
> oversaw it.

No worries.

> This file was correctly detected before (see
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1138710#71) as
> 
> images_helpimg.zip:          Zip archive data, made by v2.0 UNIX, extract 
> using at least v2.0, last modified Jun 01 2026 21:34:46, uncompressed size 
> 1810, method=store
> 
> with file 5.47 I get
> 
> images_helpimg.zip:          Microsoft OOXML

This is a regression introduced in upstream commit

| https://github.com/file/file/commit/FILE5_46-106-g8cef37b5
| commit 8cef37b568008dceaae5b6ac1ee37827310da394
| Author: Christos Zoulas <[email protected]>
| Date:   Thu May 29 15:03:04 2025 +0000
|
|     PR/0644: omajid: Recognize NuGet packages

in an attempt to detect nuget, yet another file format using .zip as a
container, and that obviously has some side effects.

The way file(1) tries to identfy what's inside a .zip file is a big
mess, I prefer to not explain this here in detail. still I'll check
whether something can be done about this.

About the # that brought you here: Did I understand correctly
strip-nondeterminism adjusts the timestamps in the file listing of a
simple .zip file - but not if it's OOXML? Then aligning that might be a
way out of your current problem. Of course, file(1) still needs to
improve here.

    Christoph

Attachment: signature.asc
Description: PGP signature

Reply via email to