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

Re: unreproducable bugs

On 7/15/05, Rich Walker <rw@shadow.org.uk> wrote:
> > I am having a hard time reading this as anything but a non sequitur.
> Umm; it follows more from Manoj's comment than yours.

Ah.  OK.

> > Personally, I prefer for a solution to be demonstrated to work, both
> > socially and technically, before it is enshrined in policy.  Drafts
> > are, of course, welcome at any stage.  "Rough consensus and running
> > code."  YMMV.
> You scale an organisation, I understand, by removing the *need* for
> everyone in it to be a genius at everything it does.

Bingo!  You also take care not to formalize unduly, or you get a
sclerotic bureaucracy.

> Hence the comment about the US army: "designed by genius to be run by
> sergeants".

As a close associate of several sergeants in the US Army, I question
only the "designed by genius" part.  Given what armies do for a
living, Darwinian selection is probably also a factor.  :-)

> There does seem to be a lot of discussion on the debian groups about
> policy. If Debian is lucky, or well-managed, then it is the process you
> are describing. If it is unlucky, then it is a bunch of rule-lawyers
> having fun.

Don't knock rule-lawyers, just ignore them until they produce
something you can tolerate.  And keep your eyes open for things that
you don't want to agree with but that happen to reflect a real-world
truth of which you were previously unaware.  Kinda like real lawyers,

> > Well, yes -- as long as the indent / emacs-mode / vim-mode
> > incantations that reproduce them are well documented, preferably in a
> > magic comment at the end of each file.  :-)
> Exactly: that and an indent script in the checkin routine remove any
> issue.

As long as it's purely advisory, please -- no tool is perfect
(although TeX is damn close).

> See how that compares to policy, which is hopefully implemented in such
> a way as to be mechanically testable?

To within certain limits, as demonstrated by lintian and linda -- up
there with dpkg and debhelper in the pantheon of Debian's
contributions to the world.  Not quite on par with the DFSG, but
that's only to be expected; the DFSG is not intended to be testable by
a machine that is less than Turing-complete.  :-)

- Michael

Reply to: