Defining specific problems and handwaving at solutions (was Re: What's Canonical thinking about Bazaar?)

[Alexander Belchenko]

Ian Clatworthy пишет:
> Let's be brutally honest ... Bazaar is *Python* and a tool by developers
> for (mostly) developers. Contributing to it *ought* to be a pleasure.
> Please step forward with your ideas on how we can make it that. How do
> other successful open source projects solve this? What can we learn from
> others?

I want to tell you some of my story.

First at all I'm totally agree with Matthew. Although it seems I still 
have commit rights but I have problems to use it. And one of the 
problem: second approve vote exists even for non-core committers. I've 
worked on Windows support very long time. Because I saw it's a free area 
and somebody have to work here. And it was always much easier for me to 
get windows-specific patches in, because there was only John who may 
block or correct me. Other devs are not really dive inro this clearly 
uninteresting stuff.

Second. Couple of years ago when I was seriously disappointed by bzr and 
all this locks nightmare I've started looking at Mercurial. 
Unfortunately I was and still is too addicted to bzr and QBzr so I'm 
still can't switch; hg has enough its own anti-windows quirks so it's 
beyond my ability to tolerate. But I digress. I want to tell about my sin.

I watched their mailing list several weeks and one time I saw 
windows-related problems with stdin/stdout and invoking hg as 
subprocess. I've recall we have fixed similar problem in bzr, so I've 
looked at bzr code and suggest similar untested solution to hg guys. I'm 
even did not expect the patch from non-member of hg community will be 
accepted. At least without boring code review and long discussion about 
pros and cons. But there happened something unbelievable for me: my 
patch was accepted and committed to hg crew branch in several hours. Can 
you imagine this? I'm still can't. I was so shocked by this fact so I've 
started to test my proposed solution, found a bug and have to send 
another "right" patch with apologies. Otherwise they can call me "saboteur".

So the sum of my story. Hg community does not have zillion tests for 
every patch and may accept patches much simpler for casual contributors. 
Hg has ugly testing subsytem which is not easy to run on Windows. And 
nevertheless hg is more popular even on Windows. Paradox? I think it's 
not really.

I've watched bzr development many years (since second half of 2005). And 
I see this every time: while hg guys are pragmatic and happy with simple 
system; bzr core devs are idealists and just can't accept 80% solution 
(as with content filters/versioned properties, or nested trees) -- 
patches should solve problems only by 100%. Or maybe 99%. No less.

The bar is too high. That's what I think.

I'm sorry to say all this, but as plain user I can't wait for decades 
until ideal 100% solution for problem that bother me will land. I need 
to get things done with my own job, and vcs should either help me or 
don't block me. Otherwise I'll think about switch to other vcs.

And I don't care either Canonical copyright holder for bzr sources, or 
Bazaar become Canonical Bazaar one day. I need the tool which helps me 
get my f** job done. That's all. I'm really wish bzr will be such tool 
sooner rather than later.

Excuse moi if I say something bad.