On Mon, Jan 26, 2015 at 2:36 PM, Tudor Girba <tudor@tudorgirba.com> wrote:
The situation is like this:
PetitParser depends on Glamour
Glamour depends on GlamourCore
GlamourCore contains Brick which is still very much changing these days
At the same time, GlamourCore already exists in the Pharo image (because of GT)

The problem is that if you load PetitParser, the newer packages of GlamourCore do not get loaded for some reason and this leads to the mentioned issues. I do not know how to solve this actually, and I could use some help.

I would like to share a recent exp. regarding debugging several configurations in Moose and the tool support available to debug these configurations.

I debugged quite a few configurations to obtain a customized, working configuration of Moose 5.0 for our tools. I heavily used and appreciated a lot, the project map tab in GTInspector to see project dependencies, and record command in Metacello + EyeInspector to see the actual packages being loaded. 

GTInspector on MetacelloMCVersion object shows the projects and packages that are loaded by a ConfigOf baseline or version. This can be obtained by inspecting: ConfigOf project version: XXX. This is quite helpful given the absence of any tool support to debug configurations (except actually loading and looking at the transcript :).

However, the above tool only helps determining what configOf loads in theory. However, in practice, there is complex machinery working behind the scene i.e. sometimes, a project is already loaded and the configuration might be trying to load an earlier version. In that case, I found record directive quite handy. This I obtained as: EyeInspector inspect: (configOfYYY project version: XXX) record loadDirective. It provides a list of actual project and package versions loaded by the configOf given the current state of the image. With the indented output of the loaded versions, one can see what gets loaded and the path to get there. It might be beneficial to obtain a project map for this too.




On Mon, Jan 26, 2015 at 2:02 PM, Johan Fabry <jfabry@dcc.uchile.cl> wrote:
Hi all,

just to notify that in today’s Pharo 40461 doing a ConfigurationOfPetitParser loadDevelopment leads to a broken image: red morphs of death appear in already open windows and opening the spotter gives a DNU GLMBrick>>widthBlock It would be cool if some knowledgeable person had a look at this, thanks :-)


---> Save our in-boxes! http://emailcharter.org <---

Johan Fabry   -   http://pleiad.cl/~jfabry
PLEIAD lab  -  Computer Science Department (DCC)  -  University of Chile

Moose-dev mailing list


"Every thing has its own flow"

Moose-dev mailing list