Hi,
I did a quick tour over the issues:
- I renamed versions default11 to default, because default should
always point to the latest development configuration, and from now on
we move development to Pharo 1.1
- I fixed the Glamour problem, and now all tests are green (with the
expected failures)
- I also fixed a couple of GroupTests
However, I noticed that the GroupTests break because the imported
model now contains 42 classes instead of 48. I have no idea why.
Cheers,
Doru
On 4 Jun 2010, at 22:04, Alexandre Bergel wrote:
Good job!!
Alexandre
On 4 Jun 2010, at 12:38, Simon Denier wrote:
So I know that some of you already started to use
Moose under Pharo
1.1, and now it is time to assess the situation.
I took the latest Pharo Core on the web 11383.
I loaded the latest Pharo Conf 1.1-beta1.10517
Then I loaded default of ConfigurationOfMoose.
After loading, the first issue I knew beforehand is that there was
a conflicting dependency between the version of MorphTreeWidget
loaded by Moose default and the one which now comes by default with
Pharo 1.1
I built two new default versions for Glamour and Moose with
Pharo1.1 by removing this dependency:
(ConfigurationOfMoose project version: 'default11') load
should load a working Moose (not tested)
Tests got something like 200 errors, however most of them seem to
have common roots (meaning fix one problem and you fix dozens of
tests). There seem to be nothing really serious, just fixing
changes in API and also compiler semantics, with strange things
around.
1) For example, the following snippet of code is valid (I didn't
even know that!) but the semantic has changed:
[ :i ] value: 1 -> returns 1 under Pharo 1.0, and nil under
Pharo 1.1
2) Also it seems that the compiler now considers _ as a message
send. Just try
MSEUtilities signatureFromSmalltalkSelector: #_
and you get a nice primitive failed. Of course the importer did not
like it :)
I just replaced _ with :=
3) fix deprecation asSortedArray -> asArray sort in Hashtable
4) One fix NOT commited in Glamour:
Message send autoDeselect: produces DNU, rewritten as
autoDeselection:
in GLMMorphicRenderer>>treeMorphFor:and:
I started to do that, I'm now at 20 failures and 42 errors.
The fun thing is that the changes are not necessarily complicated,
but the effects on tests can be weird (like above). So it's
difficult sometimes to tell what is the problem. Now failing tests
are mostly in GroupTest (some difference in importing the model,
perhaps simply due to changes in the system classes imported) and
MooseAlgos, a couple of other packages...
If someone knows how to replace the following snipper, it will
solve a lot of error tests in MooseAlgos actually.
Stream endOfStreamSignal signal
--
Simon
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
--
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel
http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
_______________________________________________
Moose-dev mailing list
Moose-dev(a)iam.unibe.ch
https://www.iam.unibe.ch/mailman/listinfo/moose-dev
"Yesterday is a fact.
Tomorrow is a possibility.
Today is a challenge."