Hi Ian,
Monday, December 17, 2001, 7:06:38 PM, you wrote:
IC> On Sun, Dec 16, 2001 at 02:52:59PM +1300, David McNab wrote:
>> I was starting to implement the non-dbr inserts into fcpputsite, but I
>> can't be bothered now.
>>
>> Someone else can do it.
>> I now wash my hands of the code entirely.
IC> David,
IC> I am sorry to hear that you have been frustrated, but from what I can
IC> tell people only had the best of intentions, it was certainly not an
IC> effort to make fcptools incompatable with the windows API (which, I
IC> agree, would be a huge error). I hope you reconsider,
IC> Ian.
I've had some time to cool off somewhat.
What 'pushed my button' was having checked in code that was building
and working on *nix and 'doze, turning my back on it a few weeks
later, and finding it wouldn't build on 'doze. Worse, there were lots
of #ifdef WINDOWS statements that simply disappeared, leaving the
windows compilation to crash because windows doesn't have some of the
headers that linux has, and vice versa.
I got particularly stunned and offended by the change made to the
Sleep routine - it got changed to:
Sleep(int secs, int nsecs)
or something like that.
Anyone who codes to windows knows that Sleep is a windows API call,
and would know that substituting another Sleep that takes 2 args would
simply not be accepted by windows. On all looks of it, it seemed that
the author of the changes had decided to concentrate on the linux
build and completely disregard the windows build.
On seeing stuff like that, it occurred to me that I was going to have
to do a windows port of code that I'd already worked hard to design
with windows in mind.
It was at that point I went postal.
Like, how would hobx feel if he cvs updated and found that someone had
replaced all the crypto with a scheme that encrypts via "XOR
0x52494141" ?
I do appreciate that people are contributing to the codebase, and that
fcptools has won a level of acceptance in the Freenet suite. For
that, I can feel that my labours have created something worthwhile.
But IMO, I don't think it's too much to ask that:
1) All the fcptools code be rendered buildable on windows msvc
(unfortunately the de-facto standard windows compiler), not just
linux and cygwin
2) No code be checked in to cvs in future unless it's verified to
build clean and work on linux and windows msvc
3) The msvc files - *.dsw and *.dsp - be kept current, so that anyone
can check out the tree and build it without having to hack project
settings.
I'm open to receiving an email saying that the codebase now
compiles and works on msvc. At that time, I'll gladly get to work on
implementing non-DBR site insertions in fcpputsite (and even FreeWeb
too for that matter).
Cheers
David
_______________________________________________
Devl mailing list
Devl at freenetproject.org
http://lists.freenetproject.org/mailman/listinfo/devl
