[squeak-dev] Daily Commit Log

commits at source.squeak.org commits at source.squeak.org
Wed Apr 20 23:55:06 UTC 2011


Changes to Trunk (http://source.squeak.org/trunk.html) in the last 24 hours:

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004537.html

Name: Kernel-nice.570
Ancestors: Kernel-dtl.569

Remove #cachedStackTop
Rationale: we can't always add, we must also remove
More: it's unsent, old, implementation is weird, a request on squeak-dev about this method raised no interest, comment is outdated, no longer true and intimidating, which is probably the main reason for such long survival.
Last chance to struggle against unawareness of a casual commiter just passed.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004538.html

Name: Tools-nice.342
Ancestors: Tools-cmm.341

Add a menu to inspect Workspace variables
Also always initialize the bindings ivar.
Remove protection against uninitialized bindings.

Note: no precaution was taken to initialize uninitialized bindings of already opened Workspace.
If you ever experience trouble while evaluating in an already opened Workspace, just select 'reset variables' in the command menu (white triangle in blue window smarties)

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004539.html

Name: Compiler-nice.207
Ancestors: Compiler-ul.206

Oops, use #canDeclareInstanceVariable where due.
This message was created on purpose, and has no reason to remain unsent.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004540.html

Name: Monticello-cmm.457
Ancestors: Monticello-kb.456

Utility methods for copying missing ancestry to a repository.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004541.html

Name: Monticello-cmm.458
Ancestors: Monticello-cmm.457

Fix for copying missing ancestry.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004542.html

Name: Monticello-bf.459
Ancestors: Monticello-cmm.458

Fix highlighting of versions in repository browser as reported by Nicolas: It got thrown off by differing versions with the same name. Now use id instead of name.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004543.html

Name: Tools-fbs.300
Ancestors: Tools-dtl.299

Baby steps to Browser index death: change "systemCategoryListIndex <op> 0" to "self selectedSystemCategoryName isNil/notNil".

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004544.html

Name: Tools-fbs.301
Ancestors: Tools-fbs.300

* Complete removal of systemCategoryListIndex, replaced by selectedSystemCategory.
* selectedSystemCategoryListIndex/selectedSystemCategoryListIndex: remain, used by Morphic, and defer to selectedSystemCategory/selectedSystemCategory:.
* selectedSystemCategoryName defers to selectedSystemCategory, and all its callers now call selectedSystemCategory.
* PackagePaneBrowser>>hasSystemCategorySelected pulled up to Browser.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004545.html

Name: Tools-fbs.302
Ancestors: Tools-fbs.301

A start on dyking out classListIndex and messageCategoryListIndex.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004546.html

Name: Tools-fbs.303
Ancestors: Tools-fbs.302

Serious bugfix, allowing the Browser's hierarchy button to work once more!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004547.html

Name: Tools-fbs.304
Ancestors: Tools-fbs.303

#selectSystemCategory: reads better, even though it means the setter method name doesn't match the instvar name.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004548.html

Name: Tools-fbs.305
Ancestors: Tools-fbs.304

#selectClassNamed: will replace most senders of #classListIndex:

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004549.html

Name: Tools-fbs.306
Ancestors: Tools-fbs.305

Don't reference classListIndex.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004550.html

Name: Tools-fbs.307
Ancestors: Tools-fbs.306

Side-by-side selectedClassName/classListIndex instvars.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004551.html

Name: Tools-fbs.308
Ancestors: Tools-fbs.307

Renamed HierarchyBrowser's classList instvar to classDisplayList - the name was confusing because this list and self classList contained _similar_ but not identical data, namely that the instvar used spaces to indicate the subclass relationships between classes. Class comment added to indicate same.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004552.html

Name: Tools-fbs.310
Ancestors: Tools-fbs.308

Reuse Browser's hierarchy-flattening: it does the same job.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004553.html

Name: Tools-fbs.311
Ancestors: Tools-fbs.310

And now HierarchyBrowser>>initHierarchyForClass: doesn't send #classListIndex:.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004554.html

Name: Tools-fbs.312
Ancestors: Tools-fbs.311

By making matching space-insensitive, we can simplify implementation and remove HierarchyBrowser's special-casing.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004555.html

Name: Tools-fbs.313
Ancestors: Tools-fbs.312

Finally we can start removing lots of classListIndex refererrs.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004556.html

Name: Tools-fbs.314
Ancestors: Tools-fbs.313

More classListIndex removals. Note that CodeHolder>>spawnHierarchy now returns the newly-created browser. This is so the relevant test can close the browser afterwards.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004557.html

Name: Tools-fbs.315
Ancestors: Tools-fbs.314

Browser's classListIndex is no more.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004558.html

Name: Tools-fbs.316
Ancestors: Tools-fbs.315

(a) #selectedSystemCategory: -> #selectSystemCategory:
(b) Remove errant reference to former classListIndex instvar.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004559.html

Name: Tools-fbs.317
Ancestors: Tools-fbs.316

Finally, remove the selector entirely.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004560.html

Name: Tools-fbs.318
Ancestors: Tools-fbs.317

The start of removing messageListIndex: added the instvar, setting it in parallel with messageListIndex in places where we have tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004561.html

Name: Tools-fbs.319
Ancestors: Tools-fbs.318

#selectMessageNamed:, and adding setters for selectedMessageName in the rest of the safe places where we directly set messageListIndex.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004562.html

Name: Tools-fbs.320
Ancestors: Tools-fbs.319

Replace references to messageListIndex in places that are either difficult/impossible to test, or are Very Dangerous (and impossible for me to test right now).

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004563.html

Name: Tools-fbs.321
Ancestors: Tools-fbs.320

This change is also dangerous, in the sense that it's not covered by a test.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004564.html

Name: Tools-fbs.322
Ancestors: Tools-fbs.321

And now we can refactor selectOriginalCategoryForCurrentMethod!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004565.html

Name: Tools-fbs.323
Ancestors: Tools-fbs.322

And make the test pass!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004566.html

Name: Tools-fbs.324
Ancestors: Tools-fbs.323

A bunch more messageListIndex reference removals. #messageListIndexOf: encapsulates messageList a bit more. This is necessary because while a Browser's messageList stores strings, a MessageSet's messageList stores MethodReferences.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004567.html

Name: Tools-fbs.325
Ancestors: Tools-fbs.324

More references to messageListIndex, and a minor classListIndex fix: use hasClassSelected.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004568.html

Name: Tools-fbs.326
Ancestors: Tools-fbs.325

More adjustments: every place that sets messageListIndex must also set selectedMessageName, and we remove one access to messageListIndex.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004569.html

Name: Tools-fbs.327
Ancestors: Tools-fbs.326

Returning the model makes it easy to control the model during (and especially after) tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004570.html

Name: Tools-fbs.328
Ancestors: Tools-fbs.327

A MessageSet may not have a class/metaclass selected when you click around the lists, unlike a Browser.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004571.html

Name: Tools-fbs.329
Ancestors: Tools-fbs.328

* In Browser: the big swapover from messageListIndex -> selectedMessageName.
* In MessageSet: pulling down functionality from Browser to shield it from the changes happening in its superclass.

Ideally, MessageSet would subclass from a new CodeHolder subclass BaseMessageSet, which would take all the methods currently in CodeHolder's 'message list' category, plus any other methods from Browser needed to support same. That's a step too far for this branch.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004572.html

Name: Tools-fbs.330
Ancestors: Tools-fbs.329

Big Switchover 2: the Getters.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004573.html

Name: Tools-fbs.331
Ancestors: Tools-fbs.330

Browser no longer references messageListIndex at all. I'll save the reshaping to the next commit (which will move it to MessageSet). Places that referenced messageListIndex pushed down to MessageSet.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004574.html

Name: Tools-fbs.332
Ancestors: Tools-fbs.331

Push messageListIndex down to MessageSet: Browser no longer needs it.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004575.html

Name: Tools-fbs.333
Ancestors: Tools-fbs.332

The start of removing messageCategoryListIndex: inserting the new instvar in parallel with the existing one.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004576.html

Name: Tools-fbs.334
Ancestors: Tools-fbs.333

More "parallel rollout" of selectedMessageCategoryName, and #messageCategoryListIndex and #messageCategoryListIndex: now internally use the name.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004577.html

Name: Tools-fbs.335
Ancestors: Tools-fbs.334

No more references to messageCategoryListIndex!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004578.html

Name: Tools-fbs.336
Ancestors: Tools-fbs.335

As of this commit Browser has NO indices, and relies entirely on symbols to remember what it's browsing.

We have lots more tests now (see the ToolsTests package) but we are far from having a comprehensive test suite for Browsers. Some actions require dummy classes, and I haven't thought through how to safely do that (I must look at Monticello's test suite), and some require major surgery before being testable.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004579.html

Name: Tools-fbs.337
Ancestors: Tools-fbs.336

When you view a class comment, and switch classes, you must still be viewing a class comment.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004580.html

Name: Tools-fbs.338
Ancestors: Tools-fbs.337

Re-add the instvar "growable", accidentally removed in 332 or 333.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004581.html

Name: Tools-fbs.339
Ancestors: Tools-fbs.338

MessageSets open when earlier versions move the messageListIndex instvar from Browser to MessageSet will break without this change, because nothing explicitly sets the messageListIndex to a sane (non-nil) value. So we lazily keep things working: when someone next clicks on a message, the instvar will be set, and everything will work.

This does mean that the existing MessageSets will lose their place, though.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004582.html

Name: ToolsTests-fbs.20
Ancestors: ToolsTests-fbs.19

More tests. FileContentsBrowser>>changeMessageCategories: is broken, and the test - while written - is an expected failure. See Mantis #7615.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004583.html

Name: ToolsTests-fbs.36
Ancestors: ToolsTests-fbs.35

More tests: test BrowserTest>>testFileOutMessageCategories.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004584.html

Name: ToolsTests-fbs.17
Ancestors: ToolsTests-fbs.16

By making matching space-insensitive, we can simplify implementation and remove HierarchyBrowser's special-casing.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004585.html

Name: ToolsTests-fbs.26
Ancestors: ToolsTests-fbs.25

MessageSetTest with functional setUp/tearDown, and one failing test.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004586.html

Name: ToolsTests-fbs.19
Ancestors: ToolsTests-fbs.18

#editSelection behaviour when selectClass:'ing.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004587.html

Name: ToolsTests-fbs.35
Ancestors: ToolsTests-fbs.34

#selectMessageNamed: and #selectedMessageName don't make much sense for MessageSets (which MethodReference did you mean to select when you were browsering all-implementors-of?) so let's avoid them in the tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004588.html

Name: ToolsTests-fbs.10
Ancestors: ToolsTests-fbs.9

Initial HierarchyBrowser tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004589.html

Name: ToolsTests-fbs.30
Ancestors: ToolsTests-fbs.29

Lots of tests! LexiconTest>>testOpenFullProtocolForClass illustrates an annoyance: ProtocolBrowser class has several constructor-like methods that don't work or make sense for a Lexicon; Browser has the same. It's like the inheritance tree is exactly backwards!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004590.html

Name: ToolsTests-fbs.12
Ancestors: ToolsTests-fbs.11

Rather use the proper methods, especially since we aim for #classListIndex: to be used solely by the ToolBuilder stuff.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004591.html

Name: ToolsTests-fbs.7
Ancestors: ToolsTests-tbn.5

Woefully inadequate test suite for Tools-fbs.300's extensive refactoring.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004592.html

Name: ToolsTests-fbs.28
Ancestors: ToolsTests-fbs.27

A bunch more messageListIndex reference removals. #messageListIndexOf: encapsulates messageList a bit more. This is necessary because while a Browser's messageList stores strings, a MessageSet's messageList stores MethodReferences.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004593.html

Name: ToolsTests-fbs.21
Ancestors: ToolsTests-fbs.20

Browser's classListIndex is no more.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004594.html

Name: ToolsTests-fbs.14
Ancestors: ToolsTests-fbs.12

More tests!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004595.html

Name: ToolsTests-fbs.37
Ancestors: ToolsTests-fbs.36

Still more Browser tests!

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004596.html

Name: ToolsTests-fbs.32
Ancestors: ToolsTests-fbs.31

ProtocolBrowser, Lexicon aren't in Tools-Browser but in Protocols-Tools, so their tests belong in the ProtocolsTests package.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004597.html

Name: ToolsTests-fbs.9
Ancestors: ToolsTests-fbs.8

More tests, and catching up to Tools-fbs.304's renamings.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004598.html

Name: ToolsTests-fbs.23
Ancestors: ToolsTests-fbs.22

The start of removing messageListIndex: added the instvar, setting it in parallel with messageListIndex in places where we have tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004599.html

Name: ToolsTests-fbs.18
Ancestors: ToolsTests-fbs.17

Finally we can start removing lots of classListIndex refererrs.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004600.html

Name: ToolsTests-fbs.16
Ancestors: ToolsTests-fbs.15

And now HierarchyBrowser>>initHierarchyForClass: doesn't send #classListIndex:.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004601.html

Name: ToolsTests-fbs.34
Ancestors: ToolsTests-fbs.33

Further MessageSet subclass tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004602.html

Name: ToolsTests-fbs.25
Ancestors: ToolsTests-fbs.24

Testing #showHomeCategory's really to test #selectOriginalCategoryForCurrentMethod.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004603.html

Name: ToolsTests-fbs.11
Ancestors: ToolsTests-fbs.10

#hasClassSelected tests.

=============================================

http://lists.squeakfoundation.org/pipermail/packages/2011-April/004604.html

Name: ToolsTests-fbs.27
Ancestors: ToolsTests-fbs.26

And make the test pass!

=============================================



More information about the Squeak-dev mailing list