Re: Rationale for removing binary data from debian source package
Am Mon, Feb 07, 2022 at 10:26:16AM +0100 schrieb Mathieu Malaterre:
> Dear Debian Med,
>
> I am trying to remember the rationale for removing binary data from a
> Debian source package.
>
> Let's consider charls:
>
> ```
> Files-Excluded: */test/conformance
> */test/jlsimage
> */test/*.raw
> */test/*.jls
> */test/*.dcm
> */test/*.ppm
> */test/MR2_UNC
> ```
This was added in commit:
commit 936c46cce7fb88a032bb7f964406c5f171575f4e
Author: Andreas Tille <tille@debian.org>
Date: Thu May 19 17:27:01 2016 +0200
New upstream version (use Files-Excluded)
If I go back to
commit 3c1373841374b5fb75c4da2d2da7345ddd513943 (HEAD)
Author: Mathieu Malaterre <malat@debian.org>
Date: Fri Nov 7 08:24:52 2014 +0000
update to 3.9.6
`git blame` uncovers:
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 43) get-orig-source: $(UPSTREAM_SRC).zip
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 44) rm -rf $(DEBIAN_SRC_DIR)
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 45) unzip -q $(UPSTREAM_SRC).zip -d $(DEBIAN_SRC_DIR)
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 46) # UNIX eol
481f011f (Mathieu Malaterre 2011-04-27 19:07:43 +0000 47) dos2unix $(DEBIAN_SRC_DIR)/*.h
481f011f (Mathieu Malaterre 2011-04-27 19:07:43 +0000 48) dos2unix $(DEBIAN_SRC_DIR)/*.c*
481f011f (Mathieu Malaterre 2011-04-27 19:07:43 +0000 49) dos2unix $(DEBIAN_SRC_DIR)/*.txt
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 50) # remove dataset to reduce source size
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 51) rm -rf $(DEBIAN_SRC_DIR)/test/conformance/
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 52) rm -rf $(DEBIAN_SRC_DIR)/test/jlsimage/
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 53) rm $(DEBIAN_SRC_DIR)/test/*.raw
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 54) rm $(DEBIAN_SRC_DIR)/test/*.jls
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 55) rm $(DEBIAN_SRC_DIR)/test/*.dcm
48689d69 (Mathieu Malaterre 2011-04-27 19:44:40 +0000 56) rm $(DEBIAN_SRC_DIR)/test/*.ppm
3784391f (Mathieu Malaterre 2011-04-27 19:37:57 +0000 57) rm $(DEBIAN_SRC_DIR)/test/MR2_UNC
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 58) GZIP="--best --no-name" tar czf $(DEBIAN_SRC_TAR) $(DEBIAN_SRC_DIR)
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 59) rm -rf $(DEBIAN_SRC_DIR)
^e1fc389 (Mathieu Malaterre 2011-04-27 16:54:06 +0000 60) rm $(UPSTREAM_SRC).zip
So if you remember why you excluded those files at that point in time it
makes sense to document this. If not just keep them in if it permits a
sensible testing.
> Long story short, upstream is relying on some of these files to run
> unit tests (please no debate). Would it be acceptable to add a portion
> of these binary data back into the source package ? If so, what should
> I pay attention to?
I do not consider image data binary without source since there are image
editing tools inside Debian. The only thing we need to pay attention is
that there are no copyrighted images (like "lena" amongst these).
> ref:
> https://salsa.debian.org/med-team/charls/-/blob/master/debian/copyright#L5-13
Long story short: I'm fine with keeping all test data.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: