Ah yes ... RPackage was trying to send an odd message at one point ... that sounds right... Dale ----- Original Message ----- | From: "Camillo Bruni" <camillobruni@gmail.com> | To: Pharo-project@lists.gforge.inria.fr | Sent: Wednesday, July 11, 2012 9:58:04 AM | Subject: Re: [Pharo-project] [Metacello] Did Pharo 2.0 Update #20200 break Metacello bootstrapping? | | The problem we face right now are related to RPackage | Somehow the registration mechanism for system events (classes|methods | added|removed|modified) | got mixed up. | | Eventually this leads to some warnings when removing methods :/ | I thought I fixed it this morning but I was wrong :D | | | | On 2012-07-11, at 18:54, Dale Henrichs wrote: | | > I was trying to work with Pharo2.0 the other day and ran into weird | > modified package issues ... Metacello-Base was involved ... I was | > actually trying to get something else accomplished and don't have | > time to debug Pharo2.0 issues, since they are not likely to be | > directly caused by Metacello... | > | > Dale | > ----- Original Message ----- | > | From: "DeNigris Sean" <sean@clipperadams.com> | > | To: metacello@googlegroups.com, | > | pharo-project@lists.gforge.inria.fr | > | Sent: Wednesday, July 11, 2012 9:47:21 AM | > | Subject: [Metacello] Did Pharo 2.0 Update #20200 break Metacello | > | bootstrapping? | > | | > | (Posting to Metacello and Pharo-project lists) | > | | > | Metacello bootstraps in #20199, but not #20200, #20201, or | > | #20202. | > | | > | Try: | > | Gofer new | > | url: 'http://ss3.gemstone.com/ss/Versionner'; | > | package: 'ConfigurationOfVersionner'; | > | load. | > | | > | (Smalltalk at: #ConfigurationOfVersionner) project development | > | load:#('Core' 'Spec'). | > | | > | When the dialog pops up about loading a package with unsaved | > | changes, | > | if you choose merge and keep the current version for all the | > | removed | > | classes and methods, it bootstraps. If you "Load" instead of | > | Merge, | > | it fails with "MessageNotUnderstood: IdentitySet>>copyAfter:" | > | | > | Sean | > | | |

On 2012-07-12, at 09:52, Henrik Sperre Johansen wrote:

On 11.07.2012 18:58, Camillo Bruni wrote:

...

The problem we face right now are related to RPackage Somehow the registration mechanism for system events (classes|methods added|removed|modified) got mixed up.

Eventually this leads to some warnings when removing methods :/ I thought I fixed it this morning but I was wrong :D I wrote yesterday: "Filtering with the standard collection methods leaves you with an IdentitySet though, so you'd have to either convert it afterwards, change the curtailing code in deliver:to: (which uses Sequenceable protocol). "

actually that's bug I introduced, but the stuff currently failing is still due to double subscriptions of the RPackageOrganizer

copyAfter: is exactly that protocol, so you'd probably want to re-review the Announcement changes in addition to the fact that RPackage ends up in that code path in the first place.

Indeed my bad :P. I quickly wrote a decent implementation that does not have to copy around stuff in the case of ifCurtailed...