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

Re: trends.debian.net updated

Sean Whitton:
> Hello,
> On Sat 04 Apr 2020 at 09:28AM +02, Lucas Nussbaum wrote:
>> Well, no, there doesn't seem to be any serious user-visible issues.
>> That's why I keep wondering whether it makes sense to just keep all this
>> technical debt around.
> It could be useful to someone, and it is not clear that it is doing us
> any harm, though, right?

Some of the technical debt is "doing harm" in the sense that we will
have work around and deprecated code that linger and slow down our work
on improving Debian.

Some of the recent improvements in debhelper was vastly complicated by
the fact that I had to deal with compat 9 and earlier that does
something considerably different.
  Even after expediting the obliteration of manual sequence control
parameters in dh, the sequence-handling in dh is difficult to understand
because of fundamental differences between compat 5-9 vs. compat 10-13.

The dpkg tooling is keeping a now 8 year old (?) work around for
packages not having a build-arch/build-indep target.  That work around
is not gratis - neither in performance nor in the cognitive burden of
improving the packaging stack.

Helmut has been continuously filing patches against patches to have them
use debhelper's dh_auto_* tooling instead of calling
./configure|cmake|... directly without setting up cross tooling
correctly for cross-builds.

Therefore, I would like us to acknowledge the fact that technical debt
is doing harm in that it has a cost for our contributors.  But at the
same time, I know it is hard to compare objectively to the cost of the
alternatives (such "janitorial uploads to fix technical debts" or
"removals").  If it was easy, we would probably have computed the
optimal trade-off and solved this very old issue long ago. :)


Reply to: