Hi Andreas,

On Sat, 2007-01-13 at 00:00 +0100, Andreas Schreiber wrote: 
>    After a discussion here, some kind of re-implementing scmbug
> in Python (with or without change of the underlying software architecture)
> is probably a nice student project. scmbug has functionality and
> (more important) generality, that our approach is missing currently.
> I'm already going to find an interested student for that...

Comparetively, what we do with Scmbug is not even remotely as imprortant
as what you seem to do at www.dlr.de. Are you sure you want to be
investing in re-implementing Scmbug in Python ? But yes, generality has
always been the goal -- the only reason Scmbug can be truly useful and
survive, the way I see it.

>   I'm wondering if you might contribute to that re-implementation?
> Either in writing code or in intellectual support?

If one supplied a patch that 100% converted Scmbug to Python, without
introducing new bugs, and if the testsuite still passes, I would
probably accept it. I would not be able to "commit" to writing code in
this effort, at least for a while, but probably could pick-up some
pieces. I would certainly be dispensing what I know in terms of the
Scmbug internals.

But pragmatically, is this going to happen ? How hard would it be ?
Wouldn't this be a process that will take e.g. 3 months ? Other issues:
the bugtracker already exists bug reports that refer to files in Perl,
and Perl routines. Some bugtracker APIs were coded in Perl. e.g. we
reused Bugzilla's Perl API, and RequestTracker's.

>    Yes, of course, duplication of effort should be avoided.

This is the main reason I'm contacting you.

> I just think that one of the major problems in many software
> developments is the poor maintainability of code, which results in
> plenty of wokr and wastes a lot of time. Me, as well as many

I cannot disagree. My poor-man's approach in dealing with this issue has
been to always reformat patches I receive, to... my coding standards.
Overall trying to keep things simple. The simpler, the better. I'm not
proud of them in any way, just glad that the code is consistent and has
seemed understandable to some, so far. I often reorganized logic into
different modules, as long as the main approach to solving the problem
seemed right. Certainly, this won't scale if/when I one day stop
maintaining this. I've seen enough Perl code from other projects to know
what a mess could be coded in Perl. 

I also can't help but think that Scmbug wouldn't exist if it hadn't
started in Perl originally, as Robert mentioned.

A general philosophical argument I've heard in the past is that a
programmer that writes unmaintainable code will write unmaintainable
code in any language. I'm a little skeptical in switching into Python in
that sense. I'd still be filtering all code regardless, and perhaps
stubbornly, maintaining the only commit access to the repository. Or at
least until http://bugzilla.mkgnu.net/show_bug.cgi?id=859 is implemented
that would permit others to develop code in their own branches, while
only one person would be merging.

> other people I know, consider Perl code as unmaintainable and
> a clean re-implementation in a language like Python safes time
> (and costs) at the end. Of course, I don't want to affront you
> and your code in particular, its just a general argumentation.

Given that I've written about 90% of this thing, that (unfortunately) I
feel responsible for keeping it afloat and working, and that I don't see
a consistent stream of patches flowing in, I would have to admit I would
feel more comfortable keeping things the way they are. If someone said
they just received funding for 3 people for over a year to take over
Scmbug and dramatically improve it, I would be glad to see it take that
direction. But I'm afraid if I don't see such a big commitment I will
still be the one responsible maintaining what's left.

I'm almost glad we haven't grown that big yet. Or perhaps this is a sign
we will and we should start preparing.

I don't know. What do other people think ?

What do you think Andreas ?


_______________________________________________
scmbug-users mailing list
[email protected]
http://lists.mkgnu.net/cgi-bin/mailman/listinfo/scmbug-users

Reply via email to