This is the mail archive of the
cygwin@cygwin.com
mailing list for the Cygwin project.
Re: Unix pipes implemented successfully
On Mon, Jan 12, 2004 at 01:14:00PM -0800, Peter A. Castro wrote:
>On Sun, 11 Jan 2004, Christopher Faylor wrote:
>>On Sun, Jan 11, 2004 at 09:57:29PM +0200, linamat@moldtelecom.md wrote:
>>>>Sorry, but no. The CVS version of cygwin already has a stub of a fifo
>>>>implementation and the device/fhandler stuff is much different than
>>>>1.5.5.
>>>
>>>But if I change the patch to be compliant with last CVS, should you
>>>change the answer?
>>
>>Sorry but it's unlikely. As I said, there is already a start of an
>>implementation for this which I am unlikely to want to throw out.
>>
>>I expect to have a working fifo implementation in a couple of releases.
>>It's the next thing I'll work on after fixing the signal/thread
>>problems in the current CVS.
>
>"in a couple of releases"? Oh, come on, Chris, have a heart for once.
>Couldn't you at least look at what he's done? You've just barely
>started while he's actually got something working.
Did you look at the cygwin code in CVS? I used the word "stub" in order
to not overstate what is there now just in case someone were to actually
inspect the source and say "But it doesn't even do X!". But, I should
have known that someone would think I'm being mean and step forward to
call me on it. So let me apologize for being unclear and restate
things: there is more than a stub. The implementation sort of works but
it was based on an assumption of the way global atoms worked which
proved to be incorrect. I intend on changing it to use new signal
functionality (now in CVS) instead but that needs some development, too.
In the meantime, I don't even have the time to do a code review on an
1891 line patch, much of which would not even apply to the current code
base, which has undergone a major device/fhandler rewrite. Perhaps you
haven't noticed, but cygwin in CVS isn't too healthy right now (my
fault) and I'm devoting a considerable amount of time to fixing that.
After that there are other patches in the queue ahead of this one, like
Pierre Humblet's tty list security mega patch.
And, on top of all of that, the patch is obviously large enough to
require an assignment so, theoretically, I shouldn't even be spending
too much time looking at it, since there are license issues. Not that
that would stop me in my tracks if it was the only consideration but
it is a factor.
And, on top of that, there is no way in the world that I am going to let
another potential destabilizing patch into 1.5.6. There are already way
too many changes in 1.5.6. I'm not going to add more. 1.5.7 would be
the release after 1.5.6. 1.5.7 is "a couple" of releases after 1.5.5.
This kind of thing is what happens when someone codes away in silence
for a long time without announcing and checkpointing their intentions.
It's a fact of life with free software. I was bitten by it myself
in the early days of cygwin. It isn't fun and I don't like being the
bearer of bad news but it is how it works.
cgf
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/