Re: [wx-dev] RE: CVS: [KH] wxWidgets/src/common dcbase.cpp, 1.37, 1.38 imagfill.cpp, 1.13, 1.14
Ryan Norton
> I don't think I've been nasty during our interchange today, just firm.
Very well...
> Let
> me be frank: I sometimes get the impression that you view wxWidgets
> CVS as a
> playground, where you can commit whatever you want regardless of the
> unforeseen consequences that it may have on others. For example, I
> investigated further your reasoning regarding wxDoFloodFill, and the
> change
> that you committed. I can easily see someone disabling wxUSE_IMAGE
> and then
> wondering why wxDC::FloodFill fails (silently, I might add). There is
> no
> logical connection that disabling wxUSE_IMAGE would break
> wxDC::FloodFill.
> At least in the present situation, the problem will occur at compile
> time;
> wxUSE_IMAGE is required to use DC functions. This is definitely the
> lesser
> of the two evils in my opinion. Is it ideal? No. Could it be improved?
> Absolutely. Was your solution the answer? I don't think it was. I'm
> certainly not the final answer, but neither are you; a discussion
> would have
> thoroughly explored these issues.
First off, you're right. Optimally you'd want just guard out
wxFloodFill with wxUSE_IMAGE. My solution was just to get it working
without wxUSE_IMAGE and go from there (I thought about this prior to
committal if someone really needs to know) - as this would break every
platform in several spots where wxFloodFill is used without
wxUSE_IMAGE...
Also, you're probably not the only one to think like that. Its really
kind of a silly myth... I take a lot of precautions when I commit and
still worry that I'll break something - although clearly I miss a
header every now and then. Maybe it comes off that way because I
commit a lot... for me its just the way I do things I guess...
> I rely on wxWidgets to make a living (as do many other here--maybe
> even you
> do too! :) ). When things break, I lose valuable time determining why
> (such
> as why suddenly compilation failed today).
Nah, the only thing I use wx for is WxWikiServer, although lately I've
been thinking about writing my own little framework for it (I hardly
use anything of it - just the socket and taskbaricon stuff, and it
grows the app size by about 800%!). It doesn't mean that I don't have
any vested interest in it, however. The sheer amount of time and
effort I spent on it should be a good indicator (plus I'm a moderator
and one of the top question answerers on the forums and maintain the
best wx faq out there...).
> Then, I spent time looking at
> the change, determining it wasn't optimal, emailing you, responding to
> your
> terse letter demanding that I, in no uncertain terms, revert the
> change,
> etc. (beginning the letter with "Geez!" didn't exactly set a very good
> tone). Your message was written such that it was out of the question
> that
> you might have been wrong.
Well, what I wanted to do was to revert the revert you made to my
change (there's a peter piper thing going on here...) and discuss it
afterwards rather then just reverting it without input [I give you
credit for at least asking though - I was asleep :)].
> Everyone slips up from time to time, but it
> seems that your commits more often than not result in followup commits
> by
> someone else or you with comments like "oops" or "forget to do X" or
> "we
> also need X" or something like that.
Yeah, I do a lot of stuff (nearly as active as say, Vadim) and am not
quite as sharp as the other core devs... although I generally fix all
of my commits (don't really recall any real mishaps in recent memory)...
I can see how noting "oops" or "forgot to do X" can worry some people
though... really, I'd look as it more that I would be glad if someone
wrote something like that because it means they went back to
triple-check it and fix it themselves rather than waiting for someone
else to stumble over it...
>
> The goal is not to commit everything the second that you write it. I
> frequently let things sit for weeks as I continue to test and explore
> the
> issue at hand, depending on its complexity. I'd rather make the
> mistake of
> letting a bug go unfixed or a new feature go unadded for a short while
> than
> commit and introduce new problems, or make a problem worse. In many
> cases,
> I realize days later that I was doing something incorrectly or there
> was a
> better way to solve the problem.
That's a good outlook on things. For me I believe I am familiar enough
with the source at hand to make judgments in shorter terms for smaller
changes. I generally tend to model my commits a bit more towards what,
say, Vadim would do...
>
> I want to stress that I harbor no lingering animosity towards you. I
> will
> confess, however, that I may harbor some sense of disappointment that a
> talented, outgoing and helpful fellow programmer is perhaps a little
> too
> eager to get things done. How's that for a backhanded compliment? :)
Pretty good! (Also accurate :)).
Ryan
Ryan Norton
> working (No, I do not want to maintain a specific wx variant for each
> application that I develop),
Yeah... it only helps compile time + library size... it was never my
intention that it would actually help executable size etc...
Ryan