Traps for rc bug fixing
ma, 2005-04-25 kello 23:40 +0200, Joerg Jaspert kirjoitti:
> Its a tool where everything is written in one file, and then something
> generates the debian/ out of it.
Argh. Speaking as someone who looks at lots of different packages while
fixing RC bugs, anything that is out of the ordinary is a snare, and
will trip people up, or cause extra work. Having taken a very brief look
at yada, I can already see that it is going to be unpleasant to work on
packages using yada. Not that it is the only such innovation: lots of
packages have traps for the unwary.
For example, if the source code isn't unpacked after "dpkg-source -x",
then there is an extra step before I can start working. Since there is
no standard for what the step should be, I have to first figure it out.
(We *really* need a new source file format that allows multiple patches
to upstream source neatly.)
If "debian/rules build" does the actual unpacking of upstream source
every time, unconditionally, and overwrites any edits I may have done,
I'm going to lose some work before I figure that out.
If I have to edit debian/packages rather than debian/control, then it's
yet another complication for me, even if it makes things simpler for the
package maintainer.
RC bug fixing becomes less efficient, and less fun, when instead of
working on the bug, I must first map a mine field. Whenever I say "I",
it also applies to the security team, of course, who are going to have
to potentially fix any package in a stable release.
Reply to: