Administrator
|
When loading TWM from the Configuration Browser, the version shown is ConfigurationOfTilingWindowManager-LaurentLaffont.28. But when I click load, ConfigurationOfTilingWindowManager-PhilippeBack.28 is what actually gets loaded, presumably from my cache, and which has no stable version for 4.0. Another strange thing to note is that ConfigurationOfTilingWindowManager-PhilippeBack.28 reports its ancestor as ConfigurationOfTilingWindowManager-PhilippeBack.28! Huh?
Is anyone else experiencing this and what should be done?
Cheers,
Sean |
Administrator
|
Bump. Okay, I fixed the ancestry for Philippe's version on my machine, but that wasn't the problem. Apparently, if MC has a version with the requested number in the cache, even from a different author, it uses that. Possible solutions: 1. Disable the cache when loading via the MetaC Browser loadConfigurationWithStable: loadStable ... Gofer new disablePackageCache; ... 2. Enforce discipline that configurations must always be merged with latest from canonical repo. Ideally, committing to canonical and copying to MetaRepos is "the right thing", but IDK how we can possibly enforce this since the contributor may not have access to the canonical repo. 3. Load with the constraint that author and version are the same as shown in the browser list item I think #1 is probably safest. After all, what if there was a "LaurentLaffont.28" in the meta repo, and another different "LaurentLaffont.28" in the cache? Issue 15977: Metacello Browser should only load from the selected repo https://pharo.fogbugz.com/default.asp?15977 Fix in inbox: SLICE-Issue-15977-Metacello-Browser-should-only-load-from-the-selected-repo-SeanDeNigris.1 - Disable the cache when loading via the MetaC Browser
Cheers,
Sean |
On 15-07-15 00:09, Sean P. DeNigris wrote:
> Sean P. DeNigris wrote > 2. Enforce discipline that configurations must always be merged with latest > from canonical repo. Ideally, committing to canonical and copying to > MetaRepos is "the right thing", but IDK how we can possibly enforce this > since the contributor may not have access to the canonical repo. For the configurations of the configuration browser we could run a ci job that checks the canonical repo to check for duplicates/unmerged versions and mail the maintainer Stephan |
In reply to this post by Sean P. DeNigris
Hi Sean, I don't know if this is the same bug, but maybe you can take a look at my solution for issue"Changes" in Monticello seems to compare to the latest in the cache I disable the MCRepositoryCache access when searching for the "closest ancestor". 2015-07-15 0:09 GMT+02:00 Sean P. DeNigris <[hidden email]>: Sean P. DeNigris wrote |
Free forum by Nabble | Edit this page |