On Fri, Apr 14, 2017 at 11:00:50PM +0100, Chris Lamb wrote:
> tags 833697 + pending
> thanks
>
> Should be fixed in Git :)
>
>
> https://anonscm.debian.org/git/reproducible/diffoscope.git/commit/?id=82fbe27785f08d3623073eb98a3a4670b02fd2dc
I confirm that using 82fbe27785f08d3623073eb98a3a4670b02fd2dc,
diffoscope no longer crashes and instead produces the output:
--- TorBrowser-6.5a1-osx64_en-US.dmg
+++ TorBrowser-6.5a1-osx64_en-US.dmg.1
│ --- TorBrowser-6.5a1-osx64_en-US.dmg-content
├── +++ TorBrowser-6.5a1-osx64_en-US.dmg.1-content
│┄ Error extracting 'TorBrowser.app/Contents/Resources/omni.ja', falling back
to binary comparison ('Truncated input file (errno=-1, retcode=-30,
archive_p=94687314358976)')
│ <extensive hexdump diff>
It is surprising to me that the error message refers to
TorBrowser.app/Contents/Resources/omni.ja (which is a zip file), because
there's no error if I extract the dmg files with 7z and run diffoscope
on the directories directly. I thought in Message #32, that the error
was because of the dmg layer, not because of an individual zip file, but
now I'm not so sure.
I tested on a few other pairs of dmg files:
https://get.videolan.org/vlc/2.2.3/macosx/vlc-2.2.3.dmg
https://get.videolan.org/vlc/2.2.4/macosx/vlc-2.2.4.dmg
https://download-installer.cdn.mozilla.net/pub/firefox/releases/52.0.2/mac/en-US/Firefox%2052.0.1.dmg
https://download-installer.cdn.mozilla.net/pub/firefox/releases/52.0.2/mac/en-US/Firefox%2052.0.2.dmg
They didn't result in any errors in the text output, though the Firefox
one wrote to stderr:
bzip2: Firefox 52.0.1.dmg: trailing garbage after EOF ignored
bzip2: Firefox 52.0.2.dmg: trailing garbage after EOF ignored
In all these cases, diffoscope showed a diff of hexdumps of the bzip2
decompression of the dmg files (not the diffs of the files inside the
ISO image).
_______________________________________________
Reproducible-builds mailing list
Reproducible-builds@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/reproducible-builds
