[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

package testing procedures (was: Take over of texinfo/info packages)



Norbert Preining <preining@logic.at> wrote:

> Hi Frank!
>
> On Mon, 03 Okt 2005, Frank Küster wrote:
>> It would be better if it was closed in the changelog.  I'm not going to
>> look at the packages before late afternoon today, so just decide how
>> on your own whether you think it's worth to spend some time on it.
>
> Ok. New version closing this bug is on the tug server.
>
> Added entry to the changelog:
>   * texinfo formats are not necessary anymore, since texi2* do not use
>     the formats, and the new teTeX will not need to generate texinfo formats.
>     Thus bug #253124 will not be fixed. (Closes: #253124)

I've run the package through my usual "test suite", and unfortunately
again a problem was revealed.  I suggest that you (or everybody else
who sends in a package for sponsoring, or submits a patch that might
interfere with maintainer scripts, dependencies and similiar stuff) do a
test like this before providing the package, so that you can detect it
yourself.  This is what I usually do:

1. pbuilder login (on sid)
2. apt-get install <package> <plus see below>
3. edit sources list so that the new package
4. apt-get update && apt-get dist-upgrade
5. apt-get remove <package> <plus see below>
6. apt-get install <package> <plus see below>
5. apt-get --purge remove <package> <plus see below>

This I do once with "<plus see below>" empty, and repeat it with
sensible values for it.  E.g. for texinfo, I tried with texinfo +
tetex-base, and next would have tried with texinfo + tetex-extra;
during the removal stages, of course you have to replace the package
name which pulls in stuff (like tetex-extra) by the package whose
removal removes stuff, i.e. tetex-base.

For tex-common (and maybe for future releases of texinfo, too) we should
do the same with teTeX and TeXlive packages.  And once in a while - or
every time during a freeze - one should additionally try dist-upgrade
from the previous stable Debian release.

So this time it turned out that texinfo's postinst script does not check
whether it is called with "remove" or with "purge", and tries to execute
update-fmtutil all the time - failing when all packages are purged
together.  

Please correct that, and run these tests for texinfo and similarly for
info. 

Besides, I suggest the following patch to changelog.  The first change
is because the bug was about *sans* serif fonts, the second is because I
had problems to understand the sentence (the race by by Henry?).  The
third hunk was suggested to me by Emacs, I guess because it is no longer
necessary. 

--- area/texinfo-4.8/debian/changelog	2005-10-03 16:08:04.000000000 +0200
+++ texinfo-4.8/debian/changelog	2005-10-03 16:55:36.028339728 +0200
@@ -8,7 +8,7 @@
     - fixes loss of pipe character in TeX verbatim mode (Closes: #181793)
     - fixes incorrect html output (Closes: #204202, #205022)
     - output correct XML (Closes: #221988)
-    - adds commands for serif fonts (Closes: #277754)
+    - adds commands for sans serif fonts (Closes: #277754)
     - fixes crash on first backspace (Closes: #259561)
     - fixes buffer overflow in 'Follow xref' (Closes: #263853)
     - fixes various navigation crashes (Closes: #265605, #293337, #308280, 
@@ -30,8 +30,9 @@
   * texinfo formats are not necessary anymore, since texi2* do not use
     the formats, and the new teTeX will not need to generate texinfo formats.
     Thus bug #253124 will not be fixed. (Closes: #253124)
-  * Include a tempfile race fix based on the one by Henry Jensen 
+  * Include a tempfile race fix based on the patch by Henry Jensen 
     (CAN-2005-3011) (Closes: #328365)
+  * Upload sponsored by Frank Küster <frank@debian.org>
 
  -- Norbert Preining <preining@logic.at>  Mon,  3 Oct 2005 13:01:02 +0200
 
@@ -456,6 +457,4 @@
 Texinfo 3.1 Debian 1 - iwj
 Initial release, completely untested.
 
-Local variables:
-mode: debian-changelog
-End:
+

Regards, Frank

-- 
Frank Küster
Inst. f. Biochemie der Univ. Zürich
Debian Developer



Reply to: