[Pharo-project] Closure fixes - test & review

Michael Roberts mike at mjr104.co.uk
Sun Sep 13 18:01:30 EDT 2009


I'm trying to integrate the latest batch of Closure fixes from Eliot
(eem).  These are listed on the tracker as 9 tasks - search under the
closure tag.  I'm looking for test & review of the changes. It is
quite tricky and likely to kill your image.

The merge / load into Pharo is challenging because we don't really
want to just drop the versions into Pharo. It would back out a number
of changes. Some trivial (comments, spacing) and some less so.  Useful
testing would include trying out your own application level code (e.g
Seaside, etc) to see if anything breaks.  useful review would be if
you are familiar with changes/fixes you have put into Pharo, you can
see if they are still there, do they still work.  if you really know
the compiler / execution machinery then of course...

I'm only publishing a 1/3 of the changes at this point, so i'm not
sure how much is expected to work in the middle, but Eliot was careful
to split it up.

0.1- start with core image >= 10449 (you need the MC changes)

0.2-
ScriptLoader new
   addRepository39ToAllPackages;
   addRepositorySqueakTrunkToAllPackages;
   addRepositoryTaskForcesToAllPackages

I have published packages to the task forces area, but the others give
you access to the source on squeak trunk and the ancestry on 3.9.

Also, if you are really keen Eliot provided CS which are on each
tracker item. There are pros and cons to trying to do this directly in
MC or using CS. If you look at the tracker for the original package
you can look at the merge directly from the Squeak trunk.

The changes are in 8 parts, and tidyup/initialisation.

Part:
1- <present>

2- Kernel-MikeRoberts-404

3-
  ---preload---
   Tools-MikeRoberts.201
   System-Support-MikeRoberts.88

  ---actual---
   Compiler-MikeRoberts.128
   CompilerTests-MikeRoberts.20
   KernelTests-MikeRoberts.149

(I have paused at 3 for this review because i break things after 4)

Notes:
1- is already in Pharo (give or take a few things)
2- is a reasonable copy
3- I have started to leave changes out. Squeak appears to have
recompiled everything at steps 3/4 but i've chosen to leave that till
9. partly because it breaks. This is why i have provided stubs of the
closure initialization.

thanks,
Mike




More information about the Pharo-dev mailing list