On 13-10-08 10:28 AM, Dimitar Zhekov wrote:
On Mon, 07 Oct 2013 12:28:49 -0700 Matthew Brush mbrush@codebrainz.ca wrote:
[...]
Any chance you could find time to review and test Geany's implementation to see if anything looks fishy?
I can simply test FiF with regex = ".", which will generate a lot of output quite fast. Without the SYNC_SPAWN patches, build.c and search.c async executions are identical (except that build checks if the returned pid after successful spawn async is > 0, but still sets the channels if it's not - I wonder why).
Can you get grep to write a lot of stuff to both stdout and stderr in the same run though? What about wrapping grep in a shell script that uses `tee` (or equivalent) to dump the output to both streams?
GTK+ 2.22 is over 3 years old by now, FWIW.
I'm using 2.22 under win~1 too, and don't think it will have any problems with async exec, but we'll see.
Yeah, just pointing it out because 3 years is a lot of code churn for a project like GTK+. There's been ~15.5 thousand commits[1] since the 2.22.0 tag :)
OT: Do you have any idea how to make g_spawn_async_with_pipes() to not open a "Command Prompt" window on Windows? My new plugin filters the document text through a subprocess very often and the popup console windows are super annoying :)
[snip]
Last, you can create a small GUI subsystem wrapper which runs your console application with CreateProcess(), and use either the creation flags or the startup information to hide or minimize it.
Hmm, this is probably the best and easiest solution, thanks for the idea!
Cheers, Matthew Brush
[1]: According to `git rev-list 2.22.0^..HEAD | wc -l`, not sure if that's the right way, my gitfu is weak.