If the optional ?varname? is supplied to tpool::wait, it returns an empty string even if there are jobs pending.
I cannot verify this. For me it seems to work correctly:
lexxsrv:nscp 1> tpool::create
lexxsrv:nscp 2> tpool::post tpool0x3a7d270 "after 100000"
lexxsrv:nscp 3> tpool::post tpool0x3a7d270 "after 50000"
lexxsrv:nscp 4> tpool::wait tpool0x3a7d270 "1 2" pending
lexxsrv:nscp 5> set pending
I get a different result:
% package require Thread
% tpool::post tpool01413310 "after 100000"
% tpool::post tpool01413310 "after 50000"
% tpool::wait tpool01413310 "1 2" pending
% set pending
I'm running ActiveTcl8.5.11 on Windows 7, RedHat 5.7, and SunOS 5.9 and get the same results.
Wierd. Then somebody broke something between 2.6.6
(the last one I made any changes to) and the 2.6.7...
This needs more examination.
So far I can see, the code from 2.6.5 and 2.6.7 did not change.
Which means that it has something to do with core 8.5.11.
We are still at 8.4 and there I cannot see any problem.
I just tried it with ActiveTcl8.4.19 and see the problem.
Then we will need to debug that part :-)
Unfortunately... I cannot reproduce this at will.
I DO get sometimes the same result but only
when I use some older compilation of the extension.
If I recompile the, say, 2.6.5 fresh, I cannot get any problem.
I will look into this in more detail but this will require time.
Are you compiling the extension yourself or do you use
OK. I am now able to reproduce it. This is a kind-of race condition
which I have to understand. This is the explanation why it is sometimes
visible and why not.
It was not (as anticipated) a race-condition. It was a regular bug.
I now need to figure out how to submit my changes using this all
new fancy SCM system that I', new at...
The fix will appear in 2.7.0 release. Thank you for reporting the bug.
Thank you for fixing it!
Fixed in 2.7
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.