On Fri, Mar 16, 2007 at 08:03:52AM -0600, Charles Cazabon wrote:
> Maybe. Go look at how some of the Linux kernel folks were talking about
> implementing "pluggable" schedulers and such.
Go look at what is the kernel organization, and how the developers implement
the possibility for the user to include a feature or not in the executable.
They do it using #ifdef.
> > Sure? Are you really sure?
> Yup. I said the same thing. I know Russ knows what he's talking about, and I
> obviously think I do, too.
Ok, so djb is a novice programmer in your opinion, he used #ifdef in his code to
check for external conditions and compile a piece of code or not.
$ cat qmail-1.03/*|grep -c '#ifdef'
140
And he didn't need to create a patching system, but a program from scratches.
> > Please clarify me how can you approach our problem (not a generic
> > programming
> > issue) with your solution.
>
> That sounds suspiciously like "please implement what I can describe but not
> implement myself".
I don't say I'm the best programmer in the world, but I already showed that my
code is not a bad code (ask people using my patches).
I never asked someone to do the job for me, I usually rewrite everything from
scratch. I came out with my project because someone came out with a similar
idea,
and I don't want to create two projects with the same goal.
> Here's what it boils down to: you've proposed an approach to a technical
> problem. Many here think, based on their experience and knowledge, that it is
> a poor, unmaintainable approach. You think it's a good approach. So the next
> step is for you to go away and implement it. Either it'll work really well,
> and you'll prove us wrong, or it'll turn out poorly, and you'll have gained
> experience and knowledge in the process. Note this does *not* involve Russ or
> anyone else here showing you how to do it.
I say something more: I've proposed an approach, explained in details and
documented
with short examples of code, and no one is still proposing a better approach
that
could be used in a real world scenario.
I'm not saying that you should write the code, just proposing an alternative
approach
(not just "your approach sucks") that fits with project goals.
I'm not asking you because I've no valid ideas, but because i want to develop
this
project for the qmail community, and so I asked the community how they want it.
I'll implement it of course, in my spare time, and if I'll find it useful I'll
release
to the community.
I'm only sad because I expect the community to react in a collaboration manner,
not
in this way: qmail "seniors" as oppositors (not proposing a way to a better
development)
and the rest of the list silently lurking.
Does qmail use #ifdef: yes
Does toasters patches use #ifdef: yes
So, my solution could be not the best, but it can increase the current status.
I asked for other approaches, but I just got "not yours" replies, and I'm sad
for that:
it seems that everything I said was wrong a priori, don't you feel? So I ask:
why?
What's wrong with me? I need information to become a better person, so please
tell me.
I've nothing else to say, I'll try to make you change your mind with the code,
I'll
do my best to create something useful, hoping to complete the project before
qmail2
release :)
As soon as I have the first set of patches ready to show my model, I'll present
you
and wait for feedback.
But I really change my mind about qmail community, sorry, even if I'll continue
to try
helping people on this list.
Bye
Fabio
|