Hi Marcus, Ben and others, how can we proceed on META issue https://pharo.fogbugz.com/f/cases/14447? Anyone with deeper knowdlege of Monticello and the integration process able to help here? It's a merging issue although nothing to merge. Thx T.
What I found is that if the package you are loading does not exist, then prior to the merge Monticello creates the "new" package - presumably so it has something to compare against. (As an aside, if you do not proceed with the merge, then that "new" package remains, which is poor behaviour.) That works with a package without tags. With tags, I think it is noticing that the "new" package has no tags. So one solution might be to also add the tags when the "new" package it created pre-merge, but that seems a bit backward. Ideally the "new" package would never be registered in the system until the merge is actually executed. I logged the following to deal with that issue as a dependency for the others. https://pharo.fogbugz.com/default.asp?14616 On Fri, Jan 23, 2015 at 8:00 PM, Torsten Bergmann <astares@gmx.de> wrote:
Hi Marcus, Ben and others,
how can we proceed on META issue https://pharo.fogbugz.com/f/cases/14447?
Anyone with deeper knowdlege of Monticello and the integration process able to help here? It's a merging issue although nothing to merge.
Thx T.
On Fri, Jan 23, 2015 at 8:00 PM, Torsten Bergmann <astares@gmx.de> wrote:
Hi Marcus, Ben and others,
how can we proceed on META issue https://pharo.fogbugz.com/f/cases/14447?
Anyone with deeper knowdlege of Monticello and the integration process able to help here? It's a merging issue although nothing to merge.
Thx T.
Actually, I discovered that the problem only occurs when you wrap the package in a slice. Given that these will be isolated single package updates, without needing to synchronise with other packages, would it be possible & pragmatic to load the packages directly without a slice, to get the issues out of the way. Or maybe a script that does the rename and then pushes the dirty packages to the repository? Now do we need to be concerned with the oldnamed package being an ancestor of the renamed ? cheers -ben
How about doing it as a post load script like this (tested on my local package-cache)... oldPackageName := 'ShoutTests'. newPackageName := 'Shout-Tests'. repository := MCRepository allSubInstances detect: [ :repo | repo location = '/Users/ben/Library/Application Support/Pharo/package-cache' ]. " 'http://smalltalkhub.com/mc/Pharo/Pharo40/main' ]." oldPackage := RPackageOrganizer default packageNamed: oldPackageName. newPackage := oldPackage renameTo: newPackageName. mcWorkingCopy := newPackage mcPackage workingCopy. uniqueNewVersion := (mcWorkingCopy uniqueVersionNameIn: repository) trimBoth. newVersion := mcWorkingCopy newVersionWithName: uniqueNewVersion message: 'Issue 14439 - ShoutTests should be Shout-Tests' in: repository. newVersion ifNotNil: [ Cursor wait showWhile: [ [ repository storeVersion: newVersion. ] ensure: [ (MCVersionInspector new version: newVersion) show ] ] ]. For reference, I munged that together from tracing NautliusMonticello>>commit:in: cheers -ben On Sat, Jan 24, 2015 at 1:33 AM, Ben Coman <btc@openinworld.com> wrote:
On Fri, Jan 23, 2015 at 8:00 PM, Torsten Bergmann <astares@gmx.de> wrote:
Hi Marcus, Ben and others,
how can we proceed on META issue https://pharo.fogbugz.com/f/cases/14447?
Anyone with deeper knowdlege of Monticello and the integration process able to help here? It's a merging issue although nothing to merge.
Thx T.
Actually, I discovered that the problem only occurs when you wrap the package in a slice. Given that these will be isolated single package updates, without needing to synchronise with other packages, would it be possible & pragmatic to load the packages directly without a slice, to get the issues out of the way. Or maybe a script that does the rename and then pushes the dirty packages to the repository?
Now do we need to be concerned with the oldnamed package being an ancestor of the renamed ?
cheers -ben
On Sat, Jan 24, 2015 at 9:47 PM, Ben Coman <btc@openinworld.com> wrote:
How about doing it as a post load script like this (tested on my local package-cache)...
oldPackageName := 'ShoutTests'. newPackageName := 'Shout-Tests'. repository := MCRepository allSubInstances detect: [ :repo | repo location = '/Users/ben/Library/Application Support/Pharo/package-cache' ]. " 'http://smalltalkhub.com/mc/Pharo/Pharo40/main' ]." oldPackage := RPackageOrganizer default packageNamed: oldPackageName. newPackage := oldPackage renameTo: newPackageName. mcWorkingCopy := newPackage mcPackage workingCopy. uniqueNewVersion := (mcWorkingCopy uniqueVersionNameIn: repository) trimBoth. newVersion := mcWorkingCopy newVersionWithName: uniqueNewVersion message: 'Issue 14439 - ShoutTests should be Shout-Tests' in: repository. newVersion ifNotNil: [ Cursor wait showWhile: [ [ repository storeVersion: newVersion. ] ensure: [ (MCVersionInspector new version: newVersion) show ] ] ].
For reference, I munged that together from tracing NautliusMonticello>>commit:in:
I've set it up ready to integrate here... https://pharo.fogbugz.com/default.asp?14439 hopefully works.
On 23 Jan 2015, at 13:00, Torsten Bergmann <astares@gmx.de> wrote:
Hi Marcus, Ben and others,
how can we proceed on META issue https://pharo.fogbugz.com/f/cases/14447?
Anyone with deeper knowdlege of Monticello and the integration process able to help here? It's a merging issue although nothing to merge.
I think there is a bug in RPackage⦠somewhere. Marcus
participants (3)
-
Ben Coman -
Marcus Denker -
Torsten Bergmann