Hans Leidekker wrote:
On Thursday 29 January 2004 01:01, Duane Clark wrote:
For some reason, a comment end is missing there. But otherwise, it seems to work fine with the one important app I use file notification on.
Which app is that?
Xilinx ISE, a program for use when designing with Xilinx chips.
"Duane" == Duane Clark dclark@akamail.com writes:
Duane> Hans Leidekker wrote: >> On Thursday 29 January 2004 01:01, Duane Clark wrote: >> >>> For some reason, a comment end is missing there. But otherwise, it >>> seems to work fine with the one important app I use file >>> notification on. >> >> >> Which app is that?
Duane> Xilinx ISE, a program for use when designing with Xilinx chips.
XST runs like a crawl, when started from the GUI. It speeds, when run from the command line. Is the behaviour of the file notification the reason for this behaviour?
Bye
Uwe Bonnes wrote:
"Duane" == Duane Clark dclark@akamail.com writes:
Duane> Hans Leidekker wrote: >> On Thursday 29 January 2004 01:01, Duane Clark wrote: >> >>> For some reason, a comment end is missing there. But otherwise, it >>> seems to work fine with the one important app I use file >>> notification on. >> >> >> Which app is that? Duane> Xilinx ISE, a program for use when designing with Xilinx chips.
XST runs like a crawl, when started from the GUI. It speeds, when run from the command line. Is the behaviour of the file notification the reason for this behaviour?
Bye
No, that is the named pipe problem. I retain a private patch in my tree, which reverts the named pipe patch that causes the crawl. With that, ISE works fine on a current CVS. It is my dream that one of these days, I'll actually have some free time to investigate how named pipes work and maybe try to fix it. I'd be happy to email the diff against the current CVS to you if you want.
With ISE, I use an external file editor (because the builtin one is junk), and ISE recognizes when I save a file with the external editor and sets up to recompile. I believe that is the file notification features for that, though I have not verified it.
On Thu, Jan 29, 2004 at 05:23:24PM -0800, Duane Clark wrote:
No, that is the named pipe problem. I retain a private patch in my tree, which reverts the named pipe patch that causes the crawl.
Just a thought: This is not the only time, where a hack is used to fix a specific problem/application that is unacceptable for inclusion into the source tree. How about adding a hacks/ directory where patches like this one can be collected and (more or less) maintained? The maintainance would be the responsibiblity of those who use the patch.
Ciao Jörg