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

Re: testing packages at build

Op do 09-10-2003, om 14:15 schreef Bill Allombert:
> My first goal is to persuade developers that running tests is
> worthwhile. For the implentation I have mainly 3 questions:
> 1) Do porters and autobuilders admins want to be able to skip the tests ?

Not me. Running regression tests is (very) good, as it allows for early
discovery of non-obvious bugs, which in turn improves the quality of our
distribution (at least if the build is made to fail if ((too) many)
tests fail).

The benefits of not running regression tests at build time (saving a
considerable amount of time) do, IMO, not outweigh the disadvantages
(having to dig deep in weird build failures with mysterious origins to
find out what the heck is making all those packages fail, if a
regression test did not catch an obsure bug).

> 2) Do we need a more featureful test machinery that just running test
> in the debian/rules build ?

I wouldn't say we need it, but I wouldn't object to having it either.

> 3) Do we want to allow for autorecovery ? If gcc -O2 leads to a broken
> binary, why not set up debian/rules to automatically retry with gcc
> -O0 ?

Now *that* would be a needless waste of time. It's far more efficient to
actually find out what the hell is wrong with a source package or with
the toolchain and fix it, rather than trying to build a package two or
more times only to fail it, so that once a human actually found out what
the hell is wrong, it can be rebuilt the (n+1)th time.

Wouter Verhelst
Debian GNU/Linux -- http://www.debian.org
Nederlandstalige Linux-documentatie -- http://nl.linux.org
If you're running Microsoft Windows, either scan your computer on
viruses, or stop wasting my bandwith and remove me from your
addressbook. *now*.

Attachment: signature.asc
Description: Dit berichtdeel is digitaal ondertekend

Reply to: