This is the mail archive of the
cygwin-developers
mailing list for the Cygwin project.
Re: 1.7.2 release?
On Tue, Jan 26, 2010 at 08:27:54PM +0100, Corinna Vinschen wrote:
>On Jan 26 14:10, Christopher Faylor wrote:
>> On Tue, Jan 26, 2010 at 07:02:13PM +0100, Corinna Vinschen wrote:
>> >Hi,
>> >
>> >
>> >I think we could release 1.7.2 really soon now. I was hoping for end
>> >of next week. However, at least the following points are still open:
>> >
>> >- The AFS FS access problem. I'm working on that one.
>> >
>> >- Discussion: Linux or Windows default charsets as Cygwin default charsets
>> > in locales without explicit charset?
>> >
>> > My POV: I'm interested in POSIX/Linux interoperability in the first
>> > place. Native Windows interop is nice, but second. And eventually
>> > there is a workaround: LANG="foo_BAR.CHARSET".
>> >
>> >- Multiple reports of git+ssh interoperability. I was hoping that
>> > Eric would have a look into that...
>> >
>> >- $ echo LOG:bananas | tee >(grep "^LOG:" > file.txt)
>> > LOG:bananas
>> > tee: /dev/fd/63: Bad file descriptor
>> >
>> > Is that a fifo problem? If so, it might not be overly pressing since
>> > it would be an extension rather than a fix, wouldn't it?
>> >
>> >Anything else?
>>
>> I have been tinkering with the fifo problems. There is a select()
>> problem right now that I haven't been able to fix.
>>
>> I completely redesigned it a couple of weekends ago and made it much
>> much worse. So I'm making progress.
>>
>> If it comes down to it, I guess this wouldn't be a regression but it
>> would be nice to have the fifo stuff working better. I'm beginning to
>> think that equating named pipes with fifos might not have been a good
>> idea. It seems like there's obvious linkage between the two but the
>> underlying model really is different.
>
>Would sockets under the hood work better for fifos, just as for AF_UNIX
>sockets?
I was looking at sockets just last weekend. I cycle through regular
pipes, named pipes, mailslots, and sockets every couple of weeks,
looking for something that I missed. I can't remember why I rejected
sockets now. Time to cycle through again.
If I could get by with a cygwin-only solution this would be much easier.
I'd just implement it all using shared memory and mutexes. But, then
someone would try to use a fifo as stdin for a windows program and SPLAT.
cgf