Hi Frank, the problem is that Pharo is a community effort, and there are so much projects around that we cannot track down all the packages (but we are working on a "validator" for metacello to ensure the packages at least load). Eventually, that provokes some incompatibilities... but if you think is just because we are disordered, take a look of what happens in other communities (python, ruby, java, all of them I know have similar problems with their products). I understand your frustration, and we are trying to improve a lot, but a correct publishing mechanism will take some time and effort to come. In the mean time, I just can ask for your understanding and collaboration. Pharo 2.0, btw, will allow shadow variables (same name of variables in different classes, in same hierarchy, or same variable name between an instance variable and a method variable), a decision that also has some consequences, but that will allow this kind of conflicts to load without problems (to actually run, that's another story :) cheers, Esteban On Jan 15, 2013, at 4:01 AM, Frank Church <vfclists@gmail.com> wrote:
Is it okay for me to say that something isn't right here?
The way I see it if something in Neptuno ie GLMTreeMorphModel is going to be in conflict with MorphTreeModel, then it shouldn't even released in the first place or clearly marked as a work in progress. MorphTreeModel appoears to be quite high up the Smalltalk environment hierarchy. Unless the change which caused the conflict was caused by introducing a method in MorphTreeModel "after" it was developed in Glamour which doesn't appear to the case.
Which raises a few questions:
1. When a new version of Pharo is released, is there a set of packages with specified versions which are supposed to work perfectly with that Pharo release?
2. When a packaged is created is the creator required to list the exact base release and the combination of packages it is tested against, eg Pharo 1.4-14557?
I am not saying this with regards to Neptuno in particular because the conflict seems to come from Glamour, there is also another "announcer" method in DBXBrowser but it appears the one in GLMTreeMorphModel is the culprit. After trying what I can truthfully describe as an amateurish attempt to refactor the "announcer" method even the Finder is broken. Basically I changed "announcer" to "announcerGLM" and the Finder breaks with MNU:False>>announce and I can't revert it.
It doesn't feel right to me that search an important element in the IDE should be broken by something lower down the hierarchy. I will try it on earlier versions of Pharo to see.
On starting with Pharo I have had to switch among different versions to get the demos working properly, but this name(space) clash is simply too jarring On 14 January 2013 19:46, Esteban Lorenzano <estebanlm@gmail.com> wrote:
yes, but that means it is present twice in the hierarchy, after loading you can check when it corresponds actually and re-add announcer in MorphTreeModel, rename the one in Neptuno... and commit the fix for Neptuno to the repository... is all community work so feel free to do it :)
Esteban
On Jan 14, 2013, at 8:39 PM, Frank Church <vfclists@gmail.com> wrote:
If "announcer" is removed from MorphTreeModel isn't that likely to break something else that depends on "announcer" being present in MorphTreeModel?
Is "announcer" something new in MorphTreeModel and is Neptuno duplicating the function or is it the other round? MorphTreeModel seems to be the UI object and it doesn't feel right to me that a database oriented tool should override something in a UI object. I think if the functionality is duplicated it may be better to rename the "announcer" in Neptuno.
On 14 January 2013 16:00, Esteban Lorenzano <estebanlm@gmail.com> wrote:
that's because somethings on morphic were added for 1.4 summer. Try removing "announcer" on MorphTreeModel, then load.
On Jan 14, 2013, at 4:46 PM, Frank Church <vfclists@gmail.com> wrote:
Has anyone managed to get Neptuno - ConfigurationOfDBXTools in Pharo 1.4?.
When I try to install Neptuno in Pharo 1.4 I get the error - 'announcer is already defined in MorphTreeModel'
---------- Forwarded message ---------- From: Frank Church <vfclists@gmail.com> Date: 14 January 2013 01:05 Subject: Problem installing Neptuno To: A friendly place where any question about pharo is welcome <pharo-users@lists.gforge.inria.fr>
When I try to install Neptuno in Pharo 1.4 I get the error - 'announcer is already defined in MorphTreeModel'
Gofer it squeaksource: 'MetacelloRepository'; package: 'ConfigurationOfDBXTools'; load.
(((Smalltalk at: #ConfigurationOfDBXTools) perform: #project) perform: #version: with: #stable) load: 'Phoseydon'.
Gofer it squeaksource: 'MetacelloRepository'; package: 'ConfigurationOfDBXBrowser'; load.
(((Smalltalk at: #ConfigurationOfDBXTools) perform: #project) perform: #version: with: #stable) load: 'Neptuno' .
The error happens at the last stage when I am installing Neptuno.
The issue appears in the DBXTalk mailing list - http://forum.world.st/Re-Pharo-project-How-to-load-Phoseydon-and-Neptuno-td4...
-- Frank Church
-- Frank Church
======================= http://devblog.brahmancreations.com