Re: Summary of current experimental tags
* Niels Thykier <niels@thykier.net>, 2013-04-07, 17:33:
* python(3)-depends-but-no-python(3)-helper 2.5.4 (Nov 2011)
- In total, some 40-46 cases
- both tags are "serious/possible" (E)
- I am considering to promote to "non-experimental".
I was going to propose to un-experimental this one, too. Unfortunately,
a few of these tags are false positive. This is because people do crazy
things like this:
WITH_PYTHON2 = $(shell test -f /usr/bin/dh_python2 && echo "--with python2")
%:
dh $@ ${WITH_PYTHON2}
Lintian has of course no way of knowing with will WITH_PYTHON2 expand
to... But perhaps in these crazy cases people should just add overrides.
(FTR, this is tracked as #659335.)
* package-contains-no-arch-dependent-files
- some 250 tags in ~125 packages
- I don't think it copes with "pkg [linux-any]" cases
(i.e. where the package is arch:any to allow architecture dependent
dependencies).
Yeah, there are some false-positives of this type:
build-essential, g++-mulitlib, ... probably more.
Other reasons to make a package arch:any:
1) arch:all packages can't have (= ${binary:Version}) dependencies on
arch:any packages.
2) Unless you mark the arch:all package MA:foreign (which is not always
possible), then it's considered as native for the purposes of dependency
resolution. This is sometimes undesirable.
- It is minor/possible (I)
- Maybe promote to non-experimental
(possibly with a s/possible/wild-guess/)
I'd rather keep it experimental, but I don't have strong feelings about
this.
* package-contains-broken-symlink
- nearly 12k tags in < 1000 packages
- people seem to symlink the "weirdest" of things.
- I am no so certain that maintaining the whitelist we are doing now
will remain feasible.
It's probably not.
- it is normal/possible (W)
Certainty is inflated here; it should be wild-guess.
- I am considering to drop it.
I think we should drop it once #699059 is fixed.
* duplicate-files
- Some 35k tags in < 1000 packages
- a lot of that is caused by doxygen (etc.)
- will possibly be preceived as a lot of busy work by maintainers
The tag helped me to greatly reduce a binary package size once, but most
of the time my reaction to it is "meh, I don't care".
- dedup.debian.net is (likely to become) a good replacement for this.
- I am considering to drop it.
I don't mind dropping it.
* embedded-pear-module
No opinion about this. :)
* shlib-calls-exit
- Some 2300 tags in < 1000 packages
- Russ suggested that this tag should be dropped[2].
- it is wishlist/possible (I)
It's been almost always true positive for me, so I'd prefer if we kept
it.
--
Jakub Wilk
Reply to: