OK, I checked that mantis 6576 is superseded by 7321. To install DelayCleanup.1.cs it is necessary to write a ScriptLoader script like: script363 Delay startTimerEventLoop. (self buildConfigurationMapFor: #('Kernel-nice.renggli.300.mcz')) load Now, last thing to fix is remove SmallInteger maxVal//2 protection and test. 2009/4/19 Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com>:
Andreas answered that the SmallInteger limit is obsolete since http://bugs.squeak.org/view.php?id=6576 which was included in 3.10
So I will check that mantis 6576+7321, and change the DelayTests>>#testBounds
Then only the PREAMBLE problem will remain. But I see this preamble can simply go in a ScriptLoader script.
2009/4/19 Nicolas Cellier <nicolas.cellier.aka.nice@gmail.com>:
I took a look at http://bugs.squeak.org/view.php?id=7321 and http://code.google.com/p/pharo/issues/detail?id=688
I started publishing PharoInbox/SLICE-DelayFixFromMantis7321-nice to integrate Andreas' changes.
Now I did integrate all but DelayCleanup.1.cs This change set does cure the failing testSemaphore in Pharo, but has two problems: 1) It requires a preamble, and I don't know how they are handled in Monticello 2) It probably does not include change about LargeInteger delays included in 3.9 (Dave kindly pointed me to http://bugs.squeak.org/view.php?id=854) 3) it makes DelayTests>>#testBounds fail
I think that the limit of (SmallInteger maxVal//2) is not necessary... if LargeInteger delay fix is in the image. I can try to fix 2) mixing Andreas and Tim fixes. What about 1) ?