Sync GsDevKit/Grease with changes in SeasideSt/Grease #16
Conversation
92-Pharo-baseline-can-be-factorized
update travis lineup for the new 3.5.0 release and update README
…ssumes-new-collection 95 Don't assume Interval >> #sorted returns a new object
add pharo8 and pharo9 to testing
added Pharo 9 to baseline
… monticello metadata)
Pharo 9 and Squeak fixes
|
@jbrichau I see that this PR is hitting the bug in 3.5.1/3.5.2 that appeared just recently (SeasideSt#105) ... since there have been earlier tests that have passed against 3.5.1/3.5.2 I'm not exactly sure where the change has been made that is causing the failure ... and I have not had the cycles to spend the time to track this problem down ...and it only seems to affect the Grease travis builds ... also it seems that the change was introduced between a month and 20 days ago .... GsDevKit/Grease tests passed 2 months ago agains 3.5.1 and stated failing 19 days ago against 3.5.1. Seaside tests failed a day ago with same error against 3.5.2 and passed 26 days ago against 3.5.0. Stiil looking a bit ... do you have any ideas what may have changed? |
|
here's a passing GsDevKit_home 3.5.1build that ran three days ago that did a build with GsDevKit/Grease master branch without a build failure.... which is now very odd ... |
|
Turns out that the failure is actually occurring while SmalltalkCI is being loaded into the stone, which implies that this is a SmalltalkCI regression/bug (see SeasideSt#105 (comment)). I think that there has been some recent work in SmalltalkCI related to improving Metacello for use with Squeak and perhaps, there was a change that blows up for GemStone 3.5.1 and beyond ... |
|
After fixing SeasideSt#105 all is well! |
For Gemstone, the following changes have occurred: