Closed estebanlm closed 7 years ago
No need to switch to metadataless format ... can't do that for Metacello anyway, since it is still supported on platforms that do not support metadataless format ... The problem is that pharo does not sort the entries for json in alphabetical order ... that is a requirement that appears to have been overlooked ...
tests are a problem right now because of problems on the file server (download of pharo images timeouts) :(
I also so mv
failures: https://travis-ci.org/Metacello/metacello/jobs/276283706#L585 or is that related to a timeout? ... Let me know when it is safe to restart the tests ...
not really. That looks more like a bug on SmalltalkCI.
Have the pharo download scripts changed recently? Odd that a failure like this would make it out of travis testing...
no, they didn't
maybe SmalltalkCI is caching the download?
the downloads are being cached, but they've been cached forever and this is the first time I've seen that failure ...
I've cleared the caches and restart a build to see what will happen ...
I got the Pharo5.0 build to run (after caches were cleared) and hit a MNU:
MessageNotUnderstood: Pragma>>methodSelector
Pragma(Object)>>doesNotUnderstand: #methodSelector
MetacelloPharo30Platform>>selectorForPragma:
[ self configuration
perform: (MetacelloPlatform current selectorForPragma: pragma)
with: self ] in MetacelloBaselineConstructor(MetacelloAbstractVersionConstructor)>>evaluatePragma: in Block: [ self configuration...
don't know when the #methodSelector method was introduced for Pragma, but it is apparently not available in Pharo 5.0 ...
restarting other crashed Pharo builds ...
yes... but we are running pharo-6.1_dev... I'm not sure it should run on Pharo 5? I still do not understand what's the policy :)
ps: how it ran? After cleaning the cache?
On 09/18/2017 10:24 AM, Esteban Lorenzano wrote:
yes... but we are running pharo-6.1_dev... I'm not sure it should /run/ on Pharo 5? I still do not understand what's the policy :)
You changed code that is common to more than the 6.1 release and the the travis tests are revealing that dependency
ps: how it ran? After cleaning the cache?
Yes I cleaned the cache and but I've been using a cache for Metacello for a very long time, so it is surprising to see the failure ... maybe smalltalkCI changed how the cache is managed? and it just needs to be cleared for all projects? .. Let's hope so ...
— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/Metacello/metacello/pull/452#issuecomment-330293522, or mute the thread https://github.com/notifications/unsubscribe-auth/AAmFTzXOEJJ9NFo1kUjBIUZEIgS3Dw89ks5sjqc6gaJpZM4PZ141.
that was a mistake, I didn't wanted to merge... I just pushed manually by mistake :( (anyway I will fix it, sorry)
@estebanlm looks like you guys broke the sorting for the filetree properties files when you switched to using STON for JSON ... you'll need to sort the entries in all of your ston sections in tonel, or you will get tons of spurious changes like you are seeing in the series of commits
I will commit something for that
... and fix tests (hopefully)
but this is a mess :p can we switch to metadataless format?