Re: [glob2-devel] Refactoring

[Kai Antweiler]

>> Dependancies are by no means evil, they are quite the opposite.
>>
>> The dependancy itself can be a pain, but by depending on the library,
>> you have saved yourself countless hours of programming, and countless
>> more hours of debugging. Plus, if you have used a reputable library,
>> you are likely getting the code from someone who knows the topic of
>> the library *better* *than* *you*, which means it may have been
>> designed better than your way.

You forgot the most important thing for glob2:
By using already existing well documented libraries we make the
code clearer and attract programmers who already worked with those
libs.

I don't consider rewriting parts of well written libs a good
programming decision - but even if I'm wrong here, there is
something I definitely know:
it is a very bad strategical decision for glob2.

I consider it the worst problem of glob2 that we only have two
persons who know the code well enough.  I believe Nuage and Steph
are spending most of there free time on glob2 - and when that stops
glob2 development will stop too.
Instead of reinventing the wheel we should rebuild (or at least document)
glob2 no matter how big the effort will be. (simply because I don't
see another way glob2 will survive the next years)

Since I don't think this mail will change anything, I will continue
to ignore parts of the code that smell like reinvented ordinary library
functions and instead care for the engine.  Bradley, I strongly advice
you to do the same.  There is enough work in glob2 for everyone ;-)
The only thing you get from glob2 is the fun working on it.  It won't
make sense to pick code parts which frustrate you and work on them -
unless you remove the thing that worries you.


I remember Steph saying that: rewriting glob2 core code isn't worth
the effort.  On the other hand the effort to rewrite new code is
reasonable !?!  I don't get it.
Please rethink the dependency paradigm.


> Dependancies are a good thing, if this project would lighten up and
> have more of them, we could have saved perhaps months of coding time

..and would make coding easier in the future.
(If you have issues using ordinary classes you can give google a chance.
 With project specific classes you don't have that choice.
 Also you can consult the documentation.)


> This is however, the wrong place to discuss this..

This mailing list is for implementation issues.
Discussing general programming paradigms might not belong here.
So my previous and the first paragraph might be as wrong in this
list as the handful of mails in this thread that damn dependancies.
But the other parts of my mail are strictly glob2  specific.


However, those who implement it decide how to do it.
(But they also should document it sufficiently.)

-- 
Kai Antweiler