[V3dot10] How to crash squeak 7149

[Edgar J. De Cleene]

El 10/8/07 7:44 AM, "Ralph Johnson" <johnson at cs.uiuc.edu> escribió:

> My fix works in the 7143 image, but not in the 7143 image with
> updates.  So, now I have to figure out what got updated.
> 
> -Ralph


As  I said in previous mail, maybe the trouble is 7146DelayTweaks-ar.cs
I attach all from 7143 so all could test.
Do a updates folder in same folder of test image.
Do Utilities applyUpdatesFromDisk.
If you move some update out of folder, not is loaded, this way you could see
with and without Andreas fix.

The red test just now (7150 without Andreas fix)

#('FloatTest>>#testNaN2' 'FontTest>>#testFallback'
'FontTest>>#testParagraphFallback' 'FontTest>>#testResetAfterEmphasized'
'LocaleTest>>#testIsFontAvailable' 'MCInitializationTest>>#testWorkingCopy'
'ReleaseTest>>#testUndeclared'
'TestIndenting>>#testNewLineStartsIndentedWhenWrapped'
'TestIndenting>>#testSetUp')

Some of this could be caused in dirty image.

Edgar

-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7149] on 6 October 2007 at 7:53 am'!
"Reporter: RalphJohnson
Summary: 0006520: recursive fonts cause crash when running tests
Description: When I try to run all tests in 3.10alpha.7105, the image often crashes.  Sometimes I can run the tests once, but the second time I run them then the system crashes.  The problem is that a TTCFont has itself as its "fallbackFont", and this causes infinite recursion in methods like #ascent.  I fixed #ascent but something else broke.

Diego Deck worked on this. He said  <<<The problem (the one I found) is fired from
LocaleTest>>testLocaleChanged. Probably the problem is that the first
round of test-runs change the default fonts, and the #localeChanged
propagation fires TTCFont>>setupDefaultFallbackFont, that changes the
receiver's fallbackFont to the default font.>>>

Diego gave me the a .cs file with two changed methods.  This solves the problem for me.
Now I can run tests over and over without any problems, though several tests fail that did not fail before.
Additional Information: 
"!



ReleaseBuilderFor3dot10 new updatePackages: 'Graphics-edc.43(42).mcd
Multilingual-edc.30(29).mcd'
!

-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7148] on 6 October 2007 at 7:40:40 am'!
"Reporter: RalphJohnson
Summary: 0006702: ChangeHooksTest has a bug
Description: Running ChangehooksTest causes a notified to pop up for each test method.  The notifier is caused by tearing down the test, because it tries to restore the old change set but fails.  It tried to set the current change set with a message to ChangeOrganization, but it should have sent it to ChangeSet.  I have a patch that fixes the test.
Additional Information: 
"!

"To follow this see
http://bugs.squeak.org/view.php?id= 6702"

ReleaseBuilderFor3dot10 new updatePackages: 'SystemChangeNotification-Tests-edc.10(9).mcd'
!

-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7147] on 6 October 2007 at 7:14:38 am'!
"Reporter: Damien Cassou
Summary: 0006583: PositionableStream>>back behavior is totally wrong
Description: The current method #back is a misconception about what a stream is. A stream contains a pointer *between* elements with past and future elements. This method considers that the pointer is *on* an element.

Attached changeset:
- renames #back to #oldBack for backward compatibility
- implements #back
- adds 4 tests.

Please consider unit tests in the changeset which verifies #back and #oldBack behavior.

Additional Information: 
"!

"To follow this see
http://bugs.squeak.org/view.php?id= 6583"

ReleaseBuilderFor3dot10 new updatePackages: 'Collections-edc.91(90).mcd
CollectionsTests-edc.75(74).mcd'
!

-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7145] on 5 October 2007 at 7:17:37 am'!
"Reporter: rr
Summary: 0006704: Parser proposes to remove temps it can not remove
Description: When accepting a method, if the parser encounters an unused temporary
variable, it offers to remove it.
However half of the time the variable is stored into, and the parser does
not remove it, instead telling that you should first remove the statement
where the variable is stored.

With this fix (basically reordering the test and the user confirmation) only
the variables which are truly unused are removed.

The code of the method is simpler too.
Additional Information: 
"!

"To follow this see
http://bugs.squeak.org/view.php?id= 6704"

ReleaseBuilderFor3dot10 new updatePackages: 'Compiler-edc.61(60).mcd'
!

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 7146DelayTweaks-ar.cs
Type: application/octet-stream
Size: 895 bytes
Desc: not available
Url : http://lists.squeakfoundation.org/pipermail/v3dot10/attachments/20071008/94ea40b2/7146DelayTweaks-ar-0001.obj
-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7143] on 4 September 2007 at 7:51 am'!
ReleaseBuilderFor3dot10 new loadLast: 'Universes'
!

-------------- next part --------------
'From Squeak3.10beta of 22 July 2007 [latest update: #7143] on 4 September 2007 at 7:48:02 am'!

!ReleaseBuilderFor3dot10 methodsFor: 'updates' stamp: 'edc 9/4/2007 07:46'!
repository 
  repository isNil ifTrue: [
		repository := MCHttpRepository
                location: 'http://source.squeakfoundation.org/310'
                user: ''
                password: ''.].
  ^repository!
]style[(10 31 170)f2b,f2,f2cred;! !

ReleaseBuilderFor3dot10 new updatePackages: 'ReleaseBuilder-edc.32(31).mcd'.!