Planet Squeak Code

SlideAction class>>isSqueakV39Image

June 18, 2018 11:12 AM

Use SlidesCollectionCtx new as target for navigation button creation scripts.

CreateButton showNextSlide openInHand.
CreateButton showPreviousSlide openInHand.

CreateButton showFirstSlide openInHand.
CreateButton showLastSlide openInHand.

June 18, 2018 09:22 AM

Name: Compiler-eem.387
Author: eem
Time: 16 June 2018, 6:15:08.755516 pm
UUID: 2a8c2667-3328-4cd4-a7c9-c98e04d6ddf2
Ancestors: Compiler-eem.386

Fix the decompiler for the full lock regime. There isn't much in the decompiler that directly refers to the method inst var (most accesses use self method, which answers the sender inst var of InstructionStream). But case statements is one of them (specifically myExits := myExits reject: [ :e | e isNil or: [ e < 0 or: [ e > method endPC ] ] ]). The bug was that doClosureCopy:copiedValues: was only setting up sender but not method and so case statements were mishandled within full blocks (e.g. in LanguageEditor>>asHtml:) becausde the wrong endPC was being used.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:18 AM

Name: Compiler-eem.386
Author: eem
Time: 16 June 2018, 5:46:30.6679 pm
UUID: b7853f0c-81c4-46cf-bc35-b572d22887b7
Ancestors: Compiler-eem.385

Fix a number of failing decompilation tests due to the last statement of a block being known to answer nil by the compiler side. The root cause was my test in BlockNode>>printStatementsOn:indent: that assumed that blocks with arguments never answered nil.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:16 AM

Name: Compiler-eem.385
Author: eem
Time: 16 June 2018, 4:29:33.44317 pm
UUID: 07c5fc8e-5463-4db6-912a-fdc0125ce7df
Ancestors: Compiler-eem.384

The scanner for spotting closure creation bytecodes in methods must identify full block closure creations.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:15 AM

Name: Compiler-eem.384
Author: eem
Time: 16 June 2018, 3:23:37.766854 pm
UUID: 79e07928-38ba-489e-a330-42bca61aac02
Ancestors: Compiler-eem.383

Oops! We need that support method.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:14 AM

Name: Compiler-eem.383
Author: eem
Time: 16 June 2018, 3:20:33.200059 pm
UUID: f72d6072-25d5-4064-8c6a-03cb9752c910
Ancestors: Compiler-eem.382

Fix decompilation in the full block regime. Whem mapping from block starts to ranges, temp names, etc, the keys are Compiledblock instance4s, not simple integer pcs, and so IdentityDictioanries must be used so as not to confuse two blocks that have the same literals and bytecodes.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:13 AM

Name: Kernel-eem.1181
Author: eem
Time: 17 June 2018, 4:48:17.581384 pm
UUID: c1e480cd-6fb8-4810-af4b-0ac2298a044a
Ancestors: Kernel-eem.1180

Move Context>>blockCopy: to deprecated (BlockContext is in deprecated). This breaks Kernel's dependency on deprecated.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:09 AM

Name: Kernel-eem.1180
Author: eem
Time: 16 June 2018, 4:25:25.529049 pm
UUID: 5c0c1399-ebc8-4f60-aa94-a11fe33553d7
Ancestors: Kernel-eem.1179

No need to scan the last two literals when updating in postCopy.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:08 AM

Name: Kernel-eem.1179
Author: eem
Time: 16 June 2018, 3:48:39.465427 pm
UUID: 6fef2fe4-bb58-4546-a532-eb7da951cc4c
Ancestors: Kernel-eem.1178

Fix copying of methods containing full blocks. After creating the copy, full blocks need to update their outerCode to the copy, otherwise they are left pointing at the original.

Symptom: debugging a doit with full blocks, as soon as one steps into a block temp names disappear because the original gets copied with temp names, but any contained block method doesn't update to refer to the copy holding temp names, remaining referring to the original that has no temp names.

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 01:00 AM

Name: Kernel-eem.1178
Author: eem
Time: 16 June 2018, 2:33:01.418518 pm
UUID: 66a0a896-ea64-4d51-8bb9-14342bad7b15
Ancestors: Kernel-ul.1177

More accurate & informative symbolic printing for pushFullClosure:numCopied:

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 12:57 AM

Name: Kernel-ul.1177
Author: ul
Time: 13 June 2018, 1:03:47.457054 pm
UUID: dc254e2c-2779-4559-a782-f47fbcebf69a
Ancestors: Kernel-ul.1176

ClassDescription changes
- move compilation logic from all compile* methods to #compile:environment:classified:withStamp:notifying:logSource:
- added #compileSilently:environment:classified:notifying:, which let's you specify the environment while compiling a method silently

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 12:56 AM

Name: System-ul.1034
Author: ul
Time: 13 June 2018, 1:24:43.197886 pm
UUID: b5a05a72-7c75-4bdb-95bc-b7d1b31eb83b
Ancestors: System-nice.1033

- added Utilities class >> #useAuthorName:during:, which works just like #useAuthorInitials:during: but with AuthorName instead of AuthorInitials

by Dave Lewis (lewis@mail.msen.com) at June 18, 2018 12:52 AM

Move Context>>blockCopy: to deprecated (BlockContext is in deprecated). This breaks Kernel's dependency on deprecated.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 11:48 PM

BlockContext is in deprecated. Add the only reference to it in Kernel (now an unsent message) to break the dependency.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 11:46 PM

Form>>hackBits: must use btyesPerBasicElement:, not bytesPerElement:, to not be confused by classes such as ShortIntegerArray which pack shorts into words.


Note that on Spur this isn't necessary now that 16-bit indexability is supported directly.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 10:49 PM

Refactor the BitmapStreamTests to use assert:equals: to ease debugging the failing tests there-in.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 09:52 PM

Robustness for when a class is reoved from under the browser's feet.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 09:22 PM

Fix the decompiler for the full lock regime. There isn't much in the decompiler that directly refers to the method inst var (most accesses use self method, which answers the sender inst var of InstructionStream). But case statements is one of them (specifically myExits := myExits reject: [ :e | e isNil or: [ e < 0 or: [ e > method endPC ] ] ]). The bug was that doClosureCopy:copiedValues: was only setting up sender but not method and so case statements were mishandled within full blocks (e.g. in LanguageEditor>>asHtml:) becausde the wrong endPC was being used.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 01:15 AM

Fix a number of failing decompilation tests due to the last statement of a block being known to answer nil by the compiler side. The root cause was my test in BlockNode>>printStatementsOn:indent: that assumed that blocks with arguments never answered nil.

by Eliot Miranda (eliot.miranda@gmail.com) at June 17, 2018 12:46 AM

The scanner for spotting closure creation bytecodes in methods must identify full block closure creations.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 11:29 PM

No need to scan the last two literals when updating in postCopy.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 11:25 PM

Fix copying of methods containing full blocks. After creating the copy, full blocks need to update their outerCode to the copy, otherwise they are left pointing at the original.

Symptom: debugging a doit with full blocks, as soon as one steps into a block temp names disappear because the original gets copied with temp names, but any contained block method doesn't update to refer to the copy holding temp names, remaining referring to the original that has no temp names.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 10:48 PM

Oops! We need that support method.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 10:23 PM

Fix decompilation in the full block regime. Whem mapping from block starts to ranges, temp names, etc, the keys are Compiledblock instance4s, not simple integer pcs, and so IdentityDictioanries must be used so as not to confuse two blocks that have the same literals and bytecodes.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 10:20 PM

Correct pc range computation for full blocks. "self method" answers the home metyhod, and hence confuses things horribly when using full blocks. This is a slip; I had added the method argument but not used it in all cases. Doh!

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 09:35 PM

More accurate & informative symbolic printing for pushFullClosure:numCopied:

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 09:33 PM

Avoid a dialog during browser test involving unaccepted changes.

by Eliot Miranda (eliot.miranda@gmail.com) at June 16, 2018 08:49 PM

An include got missing, put it back where it belongs

by Tobias Pape (tobias+squeaksource@netshed.de) at June 16, 2018 07:06 PM

Add MCMethodDefinition>>protocol, a much better message name than "category".

by Eliot Miranda (eliot.miranda@gmail.com) at June 15, 2018 08:08 PM

Provide concretePCForAbstractPC:, the inverse of abstractPCForConcretePC:. Used by the TempScopeEditor

by Eliot Miranda (eliot.miranda@gmail.com) at June 14, 2018 11:02 PM

do not nag the user about setting the author initials #4

by Levente Uzonyi (leves@elte.hu) at June 14, 2018 10:57 PM

- do not nag the user about setting the author initials #3

by Levente Uzonyi (leves@elte.hu) at June 14, 2018 10:43 PM

- do not nag the user about setting the author initials #2
- fixed LangEnvBugs >> #testIsFontAvailable
- added LocaleTest >> #testIsFontAvailable to #expectedFailures

by Levente Uzonyi (leves@elte.hu) at June 14, 2018 10:35 PM

- do not nag the user about setting the author initials

by Levente Uzonyi (leves@elte.hu) at June 14, 2018 10:21 PM

- do not nag the user about setting the author initials

by Levente Uzonyi (leves@elte.hu) at June 14, 2018 10:21 PM

Fix comment on method Morph>>saveOnFile:

SmartRefStream, not SmartReferenceStream.

by David T. Lewis (lewis@mail.msen.com) at June 14, 2018 11:41 AM

Moved Morph>>beTransparent to 'accessing'
Added guard in MorphicProject>>updateLocaleDependents to check for the presence of Etoys
Fix comment in Morph>>saveOnFile (SmartRefStream)

This change makes 'New MorphicProject' work after unloading Etoys (Preference okToReinitializeFlaps not enabled )

June 14, 2018 06:47 AM

Move Morph>>beTransparent from '*Etoys-geometry' to 'accessing' category.

June 14, 2018 06:41 AM

re-categorized #referencePositionInWorld and #referencePositionInWorld: to be with related methods.

by Chris Cunningham (cunningham.cb@gmail.com) at June 14, 2018 02:48 AM

Recategorized #referencePositionInWorld (and setter) to be with related methods.
Fixed #adjustAfter: to once again allow halo rotation of morphs to actually rotate around the rotationCenter (instead of center of the morph).

by Chris Cunningham (cunningham.cb@gmail.com) at June 14, 2018 02:45 AM

Changes to various TextConverters:
#nextPut:toStream: can not be used to initialize latin1Map and latin1Encodings anymore (because most implementations reference them), so
- moved the current implementation of #initializeLatin1MapAndEncodings to ByteTextConverter, because only those converters understand #encode:.
- TextConverter >> #initializeLatin1MapAndEncodings has become subclass responsibility
- implemented #initializeLatin1MapAndEncodings in UTF* text converters
- removed class-side #initialize implementations. Will add a new one to TextConverter once all subclasses can reinitialize their tables.
- added class side #initializeTables which is expected to initialize the various class side tables in the correct order
- postscript will reinitialize class side tables where possible (all converters except for EUCTextConverter, CNGBTextConverter, EUCJPTextConverter, EUCKRTextConverter, ShiftJISTextConverter, CompoundTextConverter, KOI8RTextConverter)
- latin1Map and latin1Encodings are not initialized lazily anymore

UTF16TextConverter will take endianness into account when writing the BOM. Same thing is probably missing from UTF32TextConverter too.

by Levente Uzonyi (leves@elte.hu) at June 13, 2018 03:08 PM

ClassDescription changes
- move compilation logic from all compile* methods to #compile:environment:classified:withStamp:notifying:logSource:
- added #compileSilently:environment:classified:notifying:, which let's you specify the environment while compiling a method silently

by Levente Uzonyi (leves@elte.hu) at June 13, 2018 12:47 PM

- set AuthorName during EPSCanvasTest to avoid nagging users about it

by Levente Uzonyi (leves@elte.hu) at June 13, 2018 11:28 AM

- added Utilities class >> #useAuthorName:during:, which works just like #useAuthorInitials:during: but with AuthorName instead of AuthorInitials

by Levente Uzonyi (leves@elte.hu) at June 13, 2018 11:25 AM

Fix so that Projects>Jump To Project submenu is readable in dark color themes (by commenting out setting of the color - default works much better).

by Chris Cunningham (cunningham.cb@gmail.com) at June 13, 2018 05:20 AM

moved #rotationCenter and #rotationCenter: to base (Morphic) package in category 'rotate scale and flex'

by Chris Cunningham (cunningham.cb@gmail.com) at June 13, 2018 04:53 AM

Recategorized rotationCenter and referencePosition (and setter versions) to more a more appropriate category.
rotationCenter is the relative position of the referencePosition point within the morph. They are related - and should live together.
Also, moved base Morph versions of these out of EToys to be with the others.

by Chris Cunningham (cunningham.cb@gmail.com) at June 13, 2018 04:49 AM

Fix comment on method Morph>>saveOnFile:

SmartRefStream, not SmartReferenceStream.

June 13, 2018 12:54 AM

- do not log class creation from ClassVarScopeTest

by Levente Uzonyi (leves@elte.hu) at June 12, 2018 11:54 PM

Run ClassVarScopeTest in its own environment so that the different tests won't interfere with each other. They were randomly failing because of an issue unrelated to class variable scopes.

by Levente Uzonyi (leves@elte.hu) at June 12, 2018 11:52 PM

Move methods
Morph rotationCenter and rotationCenter:
from 'Etoys-geometry' to 'geometry'

by David T. Lewis (lewis@mail.msen.com) at June 12, 2018 12:23 PM

Move methods Morph rotationCenter and rotationCenter: from 'Etoys-geometry' to 'geometry eToy' for consistency with other implementers in Morphic, and also as required for unloading package Etoys. See also EToys-hjh.333.

by David T. Lewis (lewis@mail.msen.com) at June 12, 2018 12:22 PM

Reworked a bit inlining for profiling.

Sweeper and Selective now rebuild the free linked lists instead of maintaining them.

Micro-optimization in sweep phases.

Compaction pause effectively decreased by -60% now on a large range of benchmarks, growing heaps from a few hundred Mb to a few Gbs.

by ClementBera (bera.clement@gmail.com) at June 12, 2018 06:56 AM

- Fix id not getting assigned when a new version of an existing class is found in the image. This fixes an annoying freeze in latest trunk under those conditions.

by Chris Muller (ma.chris.m@gmail.com) at June 12, 2018 03:19 AM

Fix of 5.2 update. Sending #isMethodContext was causing a proxy re-ification and associated MagmaSession opening at a time when we're just trying to shut them down.

by Chris Muller (ma.chris.m@gmail.com) at June 12, 2018 03:19 AM

More initializationOptions clean-ups.

by Eliot Miranda (eliot.miranda@gmail.com) at June 11, 2018 10:07 PM

Having a VMMaker-specific decompiler test and decompiler test failure collecrtor helps checking for compiler issues (e.g. just discovered one of my images has a mis-compiled SpurCompactor hierarchy).

by Eliot Miranda (eliot.miranda@gmail.com) at June 11, 2018 08:50 PM

Update some class comments.

by Eliot Miranda (eliot.miranda@gmail.com) at June 11, 2018 07:44 PM

Appendix to Kernel-mt.1172 (uniclasses)

June 11, 2018 10:27 AM

Fixed a bug when new segment was allocated in C sometimes the swizzle field used by SelectiveCompactor got corrupted. Changed assertions here and there.

Now SelectiveCompactor starts to resist serious stress benchmarks... I need to try again the 20+Gb bench and it's good to go.

by ClementBera (bera.clement@gmail.com) at June 11, 2018 08:08 AM

Move methods
Morph rotationCenter and rotationCenter:
from 'Etoys-geometry' to 'geometry'

June 11, 2018 05:47 AM

Fixes use of 'find tested item' on TestCase subclasses whose name end in 'Tests' rather than 'Test'. This opens up 'find tested item' to work on 24 additional TestCase subclasses where it would not have worked before. It should result in only one false positive (LongTestCaseTestUnderTest) which will silently fail to locate, as designed

by David T. Lewis (lewis@mail.msen.com) at June 10, 2018 11:38 PM

Fixes use of 'find tested item' on TestCase subclasses whose name end in 'Tests' rather than 'Test'. This opens up 'find tested item' to work on 24 additional TestCase subclasses where it would not have worked before. It should result in only one false positive (LongTestCaseTestUnderTest) which will silently fail to locate, as designed

June 09, 2018 07:55 PM

Deprecates a method which is very much specific for the objects tool and is not general enough to remain in Morph

by Patrick Rein (patrick.rein@gmail.com) at June 08, 2018 03:03 PM

Minor refactoring of objects tool. This moves a objects tool method from Morph into the ObjectsTool. Other very minor style fixes.

by Patrick Rein (patrick.rein@gmail.com) at June 08, 2018 03:01 PM

Adds an entry to the morph export menu to allow for easy export of postscript files. As the postscript rendering is part of MorphicExtras, I added a guard.

by Patrick Rein (patrick.rein@gmail.com) at June 08, 2018 02:42 PM

Adds convenience methods for exporting a morph as postscript according to the schema used for png export

by Patrick Rein (patrick.rein@gmail.com) at June 08, 2018 02:38 PM

empty log message

June 08, 2018 09:41 AM

empty log message

June 08, 2018 09:41 AM

empty log message

June 08, 2018 09:37 AM

empty log message

June 08, 2018 09:35 AM

empty log message

June 08, 2018 09:34 AM

empty log message

June 08, 2018 09:32 AM

empty log message

June 08, 2018 09:30 AM

Port of the 7179-basic.226publ

June 08, 2018 09:30 AM

- Fixed compilation errors and some warnings.

by ClementBera (bera.clement@gmail.com) at June 08, 2018 09:21 AM

Improved assertion (we want to catch this case)

by ClementBera (bera.clement@gmail.com) at June 08, 2018 09:00 AM

- Added assertions for free chunks like crazy.
- Fixed the double linked list scheme (now everything resists the leak checker in stress GC benchs in 64 bits, previously it worked only in 32 bits)

by ClementBera (bera.clement@gmail.com) at June 08, 2018 08:40 AM

Use Berts version of Morph>>#dismissMorph, which accounts for possible presence of Connectors.

by Chris Muller (ma.chris.m@gmail.com) at June 08, 2018 01:20 AM

ObjectPropertiesMorph>>#rebuild now part of trunk.

by Chris Muller (ma.chris.m@gmail.com) at June 08, 2018 01:20 AM

One of the original uses of Connectors is to make UML diagrams. Therefore, let NCTextMorphEditor inherit SmalltalkEditor functions, so one can browse class straight from the UML diagram.

by Chris Muller (ma.chris.m@gmail.com) at June 08, 2018 01:20 AM

Fix.

by Chris Muller (ma.chris.m@gmail.com) at June 08, 2018 01:20 AM

Eliminate a deprecation warning on Squeak.

by Eliot Miranda (eliot.miranda@gmail.com) at June 07, 2018 11:24 PM

Fix static selector mapping in SpurCompactor; it needs to handle inheritance (SpurSelectiveCompactor inherits from SpurSweeper).

Spur64BitMemoryManager>>bytesBigEnoughForPrevPointer: shouldn't use =/~= for robustness.

by Eliot Miranda (eliot.miranda@gmail.com) at June 07, 2018 11:02 PM

Initialization:
Nuke the insane initializationOptions inst var in VMClass class and replace it by a simple class variable InitializationOptions. This avoids stale initializationOptions dictionaries during initialization. The exact symptom was the validation of options in VMClass class>>initializeMiscConstants; it wanted to check SpurmemoryManager's compactorClass, but since SpurMemoryManager's initialationOptions was stale it answered a bogus class. Grasshopper, enlightenment comes slowly to some.

Consequently refactor ancilliaryClasses: to ancilliaryClasses (returning to an older style).

by Eliot Miranda (eliot.miranda@gmail.com) at June 07, 2018 10:52 PM

Fixing cases where double linked list prev pointer was not correctly updated.

Still have a bug where the first free chunk of a linked list has a prev pointer instead of 0.

by ClementBera (bera.clement@gmail.com) at June 07, 2018 07:17 PM

Added bytesBigEnoughForPrevPointer: abstraction and patch all callers to use that.

32 bits works, now fixing 64 bits.

by ClementBera (bera.clement@gmail.com) at June 07, 2018 06:02 PM

- changed the free list representation from linked list to double linked list to unlink efficiently the free chunk in compaction phases.
- Fix some simulation details
- cheated a bit some simulation of VM profiling
- Removed Pig compactor (Incompatible with the new double linked list scheme)
- a few more simulation assertion on freeTreeOverlap

I have not tested yet code generation, testing right now

by ClementBera (bera.clement@gmail.com) at June 07, 2018 02:16 PM

Spur Compaction & Slang
Add support for static resolution of very limited polymorphism, specifically that required to select between the SpurSelectiveCompactor and the SpurPlanningCompactor at run-time, using SpurSelectiveCompactor for normal GC and SpurPlanningCompactor for snapshots.

SpurHybridCompactor is the compactorClass that selects between these two, based on its planNotSelect instance variable.

CCodeGenerator gets a staticallyResolvedPolymorphicReceivers dictionary that aps variable names to classes, and hence SpurHybridCompactor maps planningCompactor to SpurPlanningCompactor and selectiveCompactor to SpurSelectiveCompactor. Mapping of the selectors happens in TMethod>>prepareMethodIn: via CCodeGenerator>>ifStaticallyResolvedPolymorphicReceiverThenUpdateSelectorIn:.

All the ugliness is in SpurCompa tor class and subclass implementations of declareCVarsIn:, where we have to map the methods themselves, mapping the API (e.g. SpurPlanningCompactor>>compact to SpurPlanningCompactor>>SPC_compact, and SpurSelectiveCompactor>>compact to SpurSelectiveCompactor>>SS_compact, because SpurSelectiveCompactor inherits from SpurSweeper and mapping things twice (once from compact to SS_compact and then subsequently from SS_compact to SSC_compact) is too tricky, whereas mapping all SpurSelectiveCompactor API methods to SS_... is much easier.

Remove the otherwise unreferenced biasForGC inst var from SpurSweeper because the Slang static resolution mechanism currently only supports name mangling of selectors, not instance variables. If and when we need this we can implement it, but for the moment KISS.

To generate a VM using SpurHybridCompactor siomply set the compactorClass SpurHybridCompactor option in the VMMaker options, e.g.

generateSqueakSpurCog64VM
"No primitives since we can use those for the Cog VM"
^VMMaker
generate: CoInterpreter
and: StackToRegisterMappingCogit
with: #(ObjectMemory Spur64BitCoMemoryManager
compactorClass SpurHybridCompactor
MULTIPLEBYTECODESETS true
bytecodeTableInitializer initializeBytecodeTableForSqueakV3PlusClosuresSistaV1Hybrid)
to: (FileDirectory default pathFromURI: self sourceTree, '/spur64src')
platformDir: (FileDirectory default pathFromURI: self sourceTree, '/platforms')
including:#()

WARNING Will Robinson! I have only tested the Slang generation and VM compilation. I jhave not checked whether this runs in the simulator yet. Yes, that's backwards, but I expect this is far enough along that getting the sdimulator to work should be easy; it's the simple bit.

by Eliot Miranda (eliot.miranda@gmail.com) at June 07, 2018 03:50 AM

Convert the B3DEnginePlugin/Squeak3D methods to paragma declarations.

by Eliot Miranda (eliot.miranda@gmail.com) at June 07, 2018 03:00 AM

Name: System-nice.1033
Author: nice
Time: 6 June 2018, 11:32:39.671798 pm
UUID: 0168c19e-6a18-4ad6-a6d8-213f48663765
Ancestors: System-ul.1032

Use the well known (x ifNil: [...]) idiom rather than the longer (x ifNotNil: [:y | y] ifNil: [...])

by Dave Lewis (lewis@mail.msen.com) at June 07, 2018 12:12 AM

Name: System-ul.1032
Author: ul
Time: 31 May 2018, 10:50:35.796719 pm
UUID: 39a8d83e-2952-4f72-8528-1687b1456771
Ancestors: System-mt.1031

Added a method to Utilities to allow overriding the value of AuthorInitials while a block is being evaluated. It can be used to prevent the user being nagged during automatic method compilation when the AuthorInitials variable is not set.

by Dave Lewis (lewis@mail.msen.com) at June 07, 2018 12:11 AM

Name: Kernel-ul.1176
Author: ul
Time: 4 June 2018, 7:28:14.840591 pm
UUID: 70ad3394-d256-4a05-91aa-b5b2937739e3
Ancestors: Kernel-ul.1175

- iterative version of Context >> #copyTo: instead of a recursive one
- optimized version of Integer >> #timesRepeat:

by Dave Lewis (lewis@mail.msen.com) at June 07, 2018 12:03 AM

Name: Kernel-ul.1175
Author: ul
Time: 1 June 2018, 8:01:13.047217 pm
UUID: 5b70dacb-2f1e-40df-85db-f457ee728ed4
Ancestors: Kernel-eem.1174

Added ClassDescription >> #compileSilently:, the missing method of the #compileSilently:* family.

by Dave Lewis (lewis@mail.msen.com) at June 07, 2018 12:02 AM

Name: Kernel-eem.1174
Author: eem
Time: 31 May 2018, 4:28:54.683734 pm
UUID: 08eb9e3c-2fe3-40e5-8733-4a95f8d9cd0b
Ancestors: Kernel-topa.1173

Fix a bug in Pragma printing; immedatey after a parse but before a method is created, a pragma's method will be nil. Printing assumed it never would be; hence havoc was caused in the debugger when looking at parse trees.

by Dave Lewis (lewis@mail.msen.com) at June 07, 2018 12:01 AM

Name: Kernel-topa.1173
Author: topa
Time: 30 May 2018, 4:54:31.858379 pm
UUID: 1327e5ed-7d9c-48d4-8ff9-3d6a41a4ce73
Ancestors: Kernel-mt.1172

SubclassResp #asBit to Boolean (All other Boolean-subclass methods are overrides anyway)

For convenience, add #asNumber/#asInteger for the most common interpretation of booleans as numbers.

by Dave Lewis (lewis@mail.msen.com) at June 06, 2018 11:59 PM

Name: Compiler-eem.382
Author: eem
Time: 31 May 2018, 5:02:26.967309 pm
UUID: 52aa994c-63da-4485-aade-2f4b72f18e06
Ancestors: Compiler-eem.381

Oops! Deal with potential recursion in xLitQuoteForTokenization.

by Dave Lewis (lewis@mail.msen.com) at June 06, 2018 11:54 PM

Use the well known (x ifNil: [...]) idiom rather than the longer (x ifNotNil: [:y | y] ifNil: [...])

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at June 06, 2018 09:32 PM

Use the well known (x ifNil: [...]) idiom rather than the longer (x ifNotNil: [:y | y] ifNil: [...])

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at June 06, 2018 09:30 PM

Fix Class>>#addInstVarNames:
(is unsent, but was trying to add the already present vars twice)

Simplify some pattern:
temporaries ifNotNil: [temporaries] ifNil: [#()]
->
temporaries ifNil: [#()]
(should be an automatic rewrite rule)

Remove some unused selectors:
- BalloonBuffer class>>#mew: (mew n'est pas mieux que new)
- BitEditor>>#locateMagnifiedView:scale: (has been moved to Rectangle class)
- Scanner>>initScanner (superseded by initialize)

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at June 06, 2018 09:22 PM

- WriteBarrier support for Dictionary, Set and Interval now! Array's still can't, but planning to try to make that happen.

by Chris Muller (ma.chris.m@gmail.com) at June 02, 2018 07:48 PM

- Finally remove the old #uncompact overrides, along with their now unused "Wb"[Class]es which are no-longer-necessary since reified Closures several years ago.
- Only generate overrides once per selector instead of every occurrence in every superclass.
- Support for Dictionary and Sets now! (Note: This means WriteBarrier now overrides #at:put: for any class which overrides it from Object).

by Chris Muller (ma.chris.m@gmail.com) at June 02, 2018 07:48 PM

Support for upcoming Squeak 5.2 release which has deprecated all the subclasses of Context, which is now concrete.

by Chris Muller (ma.chris.m@gmail.com) at June 02, 2018 07:48 PM

- Support for upcoming Squeak 5.2 release which has deprecated all the subclasses of Context, which is now concrete.
- Support classes whose names are instances of WideString.
- Removed strange legacy code in MaClassIdManager>>#ensureClassDefinitionsFor: which would sometimes try to re-set the id: of the argument, because of the WideString enhancement, above. We needed to add WideString to #minimumClasses ONLY if it wasn't already part of the database's domain model already, in which case it needs to keep that one, not have it replaced by "19" the id of WideString in the minimumClasses.
- I don't know any case where a MaClassDefinition's 'id' should allow to be changed; signal a SoftwareError if that happens and we'll go from there.

by Chris Muller (ma.chris.m@gmail.com) at June 02, 2018 07:47 PM

Cogit: Answer better closed PIC data from primitiveCollectCogCodeConstituents. i.e. scan preceding methods for references to closed PICs and store temporarily the send site's first case cache tag (mapped to a class) in the PIC's methodObject field for later harvesting. Answder am array of PIC selector followed by class, target pairs, where target is either a method or #doesNotUnderstand:.

Slang:
Avoid issuing warnings for <var: 'foo' declareC: 'extern someFunction()'> pragmas.
Extract type info from extern declarations as well as static declarations.

by Eliot Miranda (eliot.miranda@gmail.com) at June 01, 2018 10:09 PM

Collect the new format PIC data.

by Eliot Miranda (eliot.miranda@gmail.com) at June 01, 2018 09:59 PM

Make tests work even if AuthorInitials is not initialized.
- #setUp of ClassTest, ClassVarScopeTest and CompiledMethodTest use #compileSilently: instead of #compile:.
- #performTest of the above classes explicltly sets AuthorInitials to the name of the class
- CompiledMethodTest >> #testDecompile uses #compileSilently:, because testing compilation is not the goal of that test
- BehaviorTest >> #testChange temporarily sets AuthorInitials

by Levente Uzonyi (leves@elte.hu) at June 01, 2018 06:09 PM

Added a method to Utilities to allow overriding the value of AuthorInitials while a block is being evaluated. It can be used to prevent the user being nagged during automatic method compilation when the AuthorInitials variable is not set.

by Levente Uzonyi (leves@elte.hu) at June 01, 2018 06:04 PM

Added ClassDescription >> #compileSilently:, the missing method of the #compileSilently:* family.

by Levente Uzonyi (leves@elte.hu) at June 01, 2018 06:01 PM

empty log message

June 01, 2018 09:12 AM

adding missed initializeWithBase

June 01, 2018 09:12 AM

Fix small but embarassing tpyo

by tim Rowledge (tim@rowledge.org) at June 01, 2018 01:33 AM

Update some more help related methods

by tim Rowledge (tim@rowledge.org) at June 01, 2018 01:29 AM

Fix profile buffer for 64-bits (use DoubleWordAray). So for symmetry change to using WordArray in 32-bits instead of Bitmap.
Hack fix MNU cases in PICs in analyzeClosedPIC:. This is a hack because the VM is answering useless values. It simply answers #doesNotUnderstand:, whereas it could more usefully answer {theClass. #doesNotUnderstand:} or simply {theClass}. In fact, all the PIC cases should supply what the actual receiver class is, not simply the target method. We have to discuss what to do. This is in Cogit>>#createCPICData:

by Eliot Miranda (eliot.miranda@gmail.com) at June 01, 2018 12:34 AM

Oops! Deal with potential recursion in xLitQuoteForTokenization.

by Eliot Miranda (eliot.miranda@gmail.com) at June 01, 2018 12:02 AM

Fix typedScanTokens: for underscore-prefixed symbols such as #_WIN32.

by Eliot Miranda (eliot.miranda@gmail.com) at May 31, 2018 11:59 PM

Fix a bug in Pragma printing; immedatey after a parse but before a method is created, a pragma's method will be nil. Printing assumed it never would be; hence havoc was caused in the debugger when looking at parse trees.

by Eliot Miranda (eliot.miranda@gmail.com) at May 31, 2018 11:28 PM

Primer proyecto Smalltalk

by Lorenzo (lamrodriguez19@gmail.com) at May 31, 2018 11:06 PM

Moved some conversion up from TextMorph to TextPlusMorph

by Karl Ramberg (karlramberg@gmail.com) at May 31, 2018 06:38 PM

A couple of fixes for TextMorph and project loading

by Karl Ramberg (karlramberg@gmail.com) at May 31, 2018 06:31 PM

Fix Sista build which is failing due to the lack of an export of objectBytesForSlots:.

by Eliot Miranda (eliot.miranda@gmail.com) at May 31, 2018 06:01 PM

Fixes an initialization issue that occurred in Squeak 5.2alpha. See NCLineMorph >> #computeBounds, too. There are places where vertices becomes nil again after #initialize. We should hunt down those cases in the near future.

May 31, 2018 09:41 AM

Fixes event handling issues that occurred in Squeak 5.2alpha.

May 31, 2018 09:39 AM

Small compatibility fix to make older system windows load. Those windows do not user SstemWindowButton but IconicButton instances.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 31, 2018 09:05 AM

Fixed consistency checks in VMClass to succeed if no compactorClass is present.

Fixed simulation of segments to re-use correctly memory. It used to be badly broken (sometimes the simulator would think there is enough room for segment + delta, but there wasn't; other segments would just overlap each other entirely).

Added segmentOverlap assertions and debugging convenience.

Finally I have the same bug in the simulator than in the real C VM! I can debug my compactor again.

by ClementBera (bera.clement@gmail.com) at May 29, 2018 02:14 PM

- fixed syntax highlighting when a smalltalk method had temporaries declared

by Levente Uzonyi (leves@elte.hu) at May 29, 2018 11:35 AM

Fixed a bug in frameless full blocks (fetching receiver from receiver index in FullBlock and not outerContext)

Add optional read barrier for temp vector accesses
(TempVectReadBarrier Slang to C compilation settings)

Fixed bugs and improved SelectiveCompactor.

Now moving to 64 bits VMMaker.

by ClementBera (bera.clement@gmail.com) at May 29, 2018 11:30 AM

Fix runaway stack given latest BorderedMorph refactorings.

by Chris Muller (ma.chris.m@gmail.com) at May 28, 2018 09:14 PM

ClassFactoryForTestCase generates class names ending with random character sequences to significantly decrease the chance of name collisions among the generated classes.

See this thread for further explanation: http://lists.squeakfoundation.org/pipermail/squeak-dev/2018-May/198941.html

by Levente Uzonyi (leves@elte.hu) at May 28, 2018 07:56 PM

Correct class comment for DirectoryChooserDialog. Add method to provide Project load/save directory choosing.

by tim Rowledge (tim@rowledge.org) at May 28, 2018 05:51 PM

Name: System-mt.1031
Author: mt
Time: 28 May 2018, 10:51:12.13037 am
UUID: 3c6fdb2a-dba4-d140-8797-7ba73b412ba1
Ancestors: System-mt.1030

Removes hard-coded 'UserObjects' category via method call.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:22 PM

Name: System-mt.1030
Author: mt
Time: 25 May 2018, 9:27:22.087382 am
UUID: 5c12364e-1739-1d46-b1d2-3c1ec7b9c67e
Ancestors: System-mt.1029

Rename #asReference to #asCodeReference.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:21 PM

Name: System-mt.1029
Author: mt
Time: 24 May 2018, 11:33:34.473783 am
UUID: 1e5f5805-6530-4bf9-a93b-dc6bca992319
Ancestors: System-mt.1028

Fixes the bug with entering/leaving projects that change the display depth.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:20 PM

Name: System-mt.1028
Author: mt
Time: 24 May 2018, 10:35:51.133621 am
UUID: e67895c1-7fda-a446-a702-5757e36441ec
Ancestors: System-mt.1027

Adds clean-up code for distinguished null-(class/method)-references.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:19 PM

Name: System-mt.1027
Author: mt
Time: 24 May 2018, 10:30:46.580621 am
UUID: 03fbd834-afce-ba4b-89d0-989e09f085bd
Ancestors: System-kfr.1026

Harmonize method references and class references.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:18 PM

Name: Kernel-mt.1172
Author: mt
Time: 28 May 2018, 10:47:42.97237 am
UUID: 814981fd-be88-c846-a898-9cef659baa2b
Ancestors: Kernel-mt.1171

Fixes the bug where uniclasses appeared in the default environment and browser and organization.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:14 PM

Name: Kernel-mt.1171
Author: mt
Time: 25 May 2018, 9:29:12.802382 am
UUID: deb4b76c-283f-764c-958d-37e8162ea996
Ancestors: Kernel-mt.1170

Rename #isReference to #isCodeReference.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:14 PM

Name: Kernel-mt.1170
Author: mt
Time: 24 May 2018, 10:24:50.225621 am
UUID: 722cd140-f6f2-8642-be85-61be05c77c92
Ancestors: Kernel-mt.1169

Harmonize method reference and class references.

by Dave Lewis (lewis@mail.msen.com) at May 28, 2018 04:13 PM

Removes hard-coded 'UserObjects' category via method call.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2018 08:51 AM

Appendix to Kernel-mt.1172 (uniclasses)

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2018 08:49 AM

New test for uniclass creation. Appendix to Kernel-mt.1172

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2018 08:48 AM

Fixes the bug where uniclasses appeared in the default environment and browser and organization.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2018 08:47 AM

Be more precise and only clean-up uni-class-hickups in environments.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2018 06:41 AM

additional methods:

SlideAction class>>

addTitleBoxWithTitleText: aText
createSlideWithTitleBox: aString content: a2ndString tags: a3rdString
createSlideWithTitleBoxText: aText content: a2ndString tags: a3rdString

May 28, 2018 05:10 AM

Adds code to clean-up environment hick-ups.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2018 11:56 AM

For clarity, only use one environment per system code browser. There is code in Browser etc. that relies on that fact.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2018 11:29 AM

Player had a send to deprecated Smalltalk>>removeKey:ifAbsent:

by Karl Ramberg (karlramberg@gmail.com) at May 26, 2018 09:55 AM

slow conversion to use 'https://github.com/cheery/lever/tree/master/lib/webassembly-1.json'

May 26, 2018 08:46 AM

Fixes the failing UTF-16 bug due to a missing initialization of the latin1 map.
Also adds validation for overlong sequences in UTF-8. Refactors some of the UTF-8 conversion code to make the bitmasks more obvious. The performance hit from the validation seems to be negligible but further testing is required.

by Patrick Rein (patrick.rein@gmail.com) at May 26, 2018 08:01 AM

- let PG3ShoutParser highlight regular smalltalk methods as its superclass would

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 09:50 PM

- fixed a typo

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 04:42 PM

stable v23

May 25, 2018 03:56 PM

- assume that rowClass of PG3Connection is always initialized when executing a query

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 10:53 AM

- always specify rowClass for refcursor functions
- postscript will recompile the methods in schema mirrors
- various tweaks for query creation and execution

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 10:53 AM

Add class comments

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 10:31 AM

- always specify the rowClass explicitly in PG3ConnectionPool >> #executeQuery:
- minor tweaks to PG3ConnectionPool >> #executeTransaction:
- merged with PostgresV3-Pool-FrancoisStephany.3

by Levente Uzonyi (leves@elte.hu) at May 25, 2018 10:30 AM

Updates the commit-message file-out code to be used as helper for writing the release notes.

Just call "ReleaseBuilder fileOutChangesBetweenReleases" and look for a new "*.md" file in your working directory.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 25, 2018 08:06 AM

Rename #isReference to #isCodeReference.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 25, 2018 07:29 AM

Rename #asReference to #asCodeReference.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 25, 2018 07:27 AM

Remove unnecessary period from generated setter methods, and blank lines from getters and setters.

by Chris Muller (ma.chris.m@gmail.com) at May 24, 2018 09:43 PM

for comodity switch to Ladrillos
Add JSON as dependency and methods for it in Object class

May 24, 2018 04:45 PM

Fixes the flickering of the text cursor over clickable text actions (such as TextURL).

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 24, 2018 03:33 PM

Fixes a hick-up with existing morphic content in the CI process.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 24, 2018 01:50 PM

Changes the Trunk version to 5.2alpha.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 24, 2018 12:54 PM

empty log message

May 24, 2018 12:39 PM

A few helpful extentions (in line with what was alredy present).
Made sure it works in 4.5 as well.

by Chris Cunningham (csque0@lycos.com) at May 22, 2018 09:51 PM

Provide ProtoObject>>basicInstVarAt: to call primitive 73 in the case of tracing a MCPackageInEnvironment instance that otherwise would delegate instVarAt: to the MCPackage for which it is a proxy. Implement this in ProtoObject rather than MCPackageInEnvironment because the image being traced my not have class MCPackageInEnvironment.

Initialize the PluggableDictionary in newLargeIdentityDictionaryOfSize: to use blocks for the hashBlock and equalBlock. This prevents problems when tracing an instance of MCPackageInEnvironment which otherwise would delegate #== to its MCPackage and is also the preferred way to initialize a PluggableDictionary.

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 11:59 PM

Re-save SystemTracing-dtl.27 as SystemTracing.V3

Explanation: SystemTracing-tfel.42.mcz incorporates support for Spur image format, and is decended from SystemTracing-dtl.27, which was the last version for V3 images. In order to permit further development of the V3 branch, save it again under the name SystemTracing.V3.

For a V3 image that follows the Squeak trunk updates, see http://www.squeaksource.com/TrunkUpdateStreamV3. Tracing that V3 image to 64 bits with SystemTracer64 will require updates to the tracer, hence this SystemTracing.V3 branch.

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 11:59 PM

Name: Collections-ul.796
Author: ul
Time: 18 May 2018, 12:37:42.876883 pm
UUID: f657f34e-d277-4433-b106-f661ef3eef1e
Ancestors: Collections-mt.795

- added missing return to ByteCharacterSet class >> #fromMap:

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 03:34 PM

Name: Collections-mt.795
Author: mt
Time: 17 May 2018, 8:08:41.389712 am
UUID: ce581475-faa1-3446-978c-84ef85a10d63
Ancestors: Collections-nice.794

Fixes an endless-loop bug as well as adds support for hrefs without quotes in the HtmlReadWriter.

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 03:33 PM

Name: Kernel-mt.1169
Author: mt
Time: 18 May 2018, 8:55:01.039917 am
UUID: e5e24186-05f2-2849-adb0-b2b2572b0f79
Ancestors: Kernel-nice.1168

Reduces the use of #cmdGesturesEnabled, which is also quite Morphic-specific. In preparation of two new preferences #haloForAll and #metaMenuForAll.

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 03:30 PM

Name: System-kfr.1026
Author: kfr
Time: 17 May 2018, 10:55:54.266357 am
UUID: ba06ecc5-87d4-3640-ae2b-e621230bfc82
Ancestors: System-mt.1025

Convert to EventSensor for older projects

by Dave Lewis (lewis@mail.msen.com) at May 20, 2018 03:28 PM

new stable v 2.9.2 (as synced May 2018 to Pharo 7 dev)

May 18, 2018 11:52 AM

Avant l'integration de Roassal2

by guemkam (guemkamy@gmail.com) at May 17, 2018 08:47 PM

empty log message

by Mortimer Cladwell (mbcladwell@stihie.net) at May 17, 2018 05:01 PM

empty log message

May 17, 2018 08:03 AM

Name: System-mt.1025
Author: mt
Time: 9 May 2018, 9:23:01.362554 am
UUID: 061ddb0a-c159-884b-8964-3f3c2f74e7c7
Ancestors: System-mt.1024

Fixes typo.

by Dave Lewis (lewis@mail.msen.com) at May 12, 2018 12:29 AM

Name: System-mt.1024
Author: mt
Time: 9 May 2018, 8:36:00.223271 am
UUID: 74d2ef0a-9d9a-2847-a108-68482bd91da4
Ancestors: System-mt.1023

Updates code comments and some code formatting.

by Dave Lewis (lewis@mail.msen.com) at May 12, 2018 12:27 AM

Name: System-mt.1023
Author: mt
Time: 6 May 2018, 3:07:01.547836 pm
UUID: 933c0495-6c3e-3947-9be9-c461397fe4e5
Ancestors: System-kfr.1022

Supplement to refactoring in Morphic-mt.1427.

by Dave Lewis (lewis@mail.msen.com) at May 12, 2018 12:25 AM

Name: System-kfr.1022
Author: kfr
Time: 3 May 2018, 8:03:36.590229 pm
UUID: 53fd1f63-34eb-f24e-ac23-99850d5b6f09
Ancestors: System-eem.1021

Class needed for bringing in old projects.

by Dave Lewis (lewis@mail.msen.com) at May 11, 2018 11:49 PM

No changes required for V3 from System-eem.1021, but clean up by moving copySmartRootsExport: back to ImageSegment, including the changes for context refactorings.
Remove all methods from NativeImageSegment but retain the class as a stub because it is referenced from other packages for Spur.

Name: System-eem.1021
Author: eem
Time: 2 May 2018, 1:39:55.407326 pm
UUID: 727d275e-a337-4208-9616-f887d8fd4576
Ancestors: System-eem.1020

Move more NativeImageSegment methods up to ImageSegment. Provide a hack accessor for testing that a segment can be loaded (forFile:outPointers:, which may get deleted once 64-bit native image segments work).

by Dave Lewis (lewis@mail.msen.com) at May 11, 2018 11:47 PM

empty log message

by Mortimer Cladwell (mbcladwell@stihie.net) at May 10, 2018 08:11 PM

First commit.

by Karl Ramberg 2 (karlramberg@gmail.com) at May 01, 2018 11:04 AM

empty log message

by Mortimer Cladwell (mbcladwell@stihie.net) at April 30, 2018 05:55 PM

PM -> PS -> Plate/well working

by Mortimer Cladwell (mbcladwell@stihie.net) at April 30, 2018 02:16 PM

CommandShell 4.7.1
Rename all menu items and window labels to refer to 'Command Shell' rather than 'Squeak Shell'. This prevents name a conflict with SqueakShellProject, and is more appropriate for Cuis and Pharo users.

by Dave Lewis (lewis@mail.msen.com) at April 30, 2018 02:08 PM

CommandShell 4.7.1
Rename all menu items and window labels to refer to 'Command Shell' rather than 'Squeak Shell'. This prevents name a conflict with SqueakShellProject, and is more appropriate for Cuis and Pharo users.

by Dave Lewis (lewis@mail.msen.com) at April 30, 2018 02:08 PM

CommandShell 4.7.1
Rename all menu items and window labels to refer to 'Command Shell' rather than 'Squeak Shell'. This prevents name a conflict with SqueakShellProject, and is more appropriate for Cuis and Pharo users.

by Dave Lewis (lewis@mail.msen.com) at April 30, 2018 02:08 PM

CommandShell 4.7.1
Rename all menu items and window labels to refer to 'Command Shell' rather than 'Squeak Shell'. This prevents name a conflict with SqueakShellProject, and is more appropriate for Cuis and Pharo users.

by Dave Lewis (lewis@mail.msen.com) at April 30, 2018 02:08 PM

empty log message

by Mortimer Cladwell (mbcladwell@stihie.net) at April 29, 2018 07:25 PM

Name: Kernel-nice.1168
Author: nice
Time: 26 April 2018, 11:40:17.923908 am
UUID: fa008d75-249a-b243-8b1e-068eeb72a45d
Ancestors: Kernel-nice.1167

Change asApproximateFraction to allow prescribing a relative decimal precision

Example:
(FloatArray with: 1/3) first
asApproximateFractionRelativeDecimalPlaces: 6.
0.6667 asApproximateFractionRelativeDecimalPlaces: 3.
0.6667 asApproximateFractionRelativeDecimalPlaces: 5.

by Dave Lewis (lewis@mail.msen.com) at April 28, 2018 08:13 PM

Name: Kernel-nice.1167
Author: nice
Time: 25 April 2018, 11:09:01.928037 am
UUID: fd3a4e4a-925e-1842-9c73-f7f5736dd21c
Ancestors: Kernel-eem.1166

I shall not invoke super floorLog: with a different radix.

Ouch, probably a remnant from my ancestor method floorLog10 that I generalized.

by Dave Lewis (lewis@mail.msen.com) at April 28, 2018 08:11 PM

Name: Kernel-eem.1166
Author: eem
Time: 24 April 2018, 11:25:52.062506 am
UUID: e2549724-d2f2-41f4-9582-6db17efefcf0
Ancestors: Kernel-eem.1165

Revert the previous convenience. It isn't. There is a better way; see MonticelloConfigurations-eem.153.

Implement FullBlockClosure>>size. (Bizarrely BlockClosure>>sized answers the numer of bytes of bytecode in the block; I don't think this is useful :-) )

by Dave Lewis (lewis@mail.msen.com) at April 28, 2018 08:10 PM

Name: Kernel-eem.1165
Author: eem
Time: 24 April 2018, 11:19:16.734066 am
UUID: 281c4af1-ffb9-4616-a9ef-a6369143c7b7
Ancestors: Kernel-eem.1162, Kernel-eem.1164

Add a convenience to turn off showing deprecation warnings for the duration of a block.

by Dave Lewis (lewis@mail.msen.com) at April 28, 2018 08:07 PM

Name: Collections-bf.787
Author: bf
Time: 26 April 2018, 1:25:29.283438 pm
UUID: ed53996c-b8a6-40b7-a186-37ec55737b62
Ancestors: Collections-pre.786

Allow to write Strings to binary WriteStream.

by Dave Lewis (lewis@mail.msen.com) at April 28, 2018 08:01 PM

NetNameResolver localHostName answers '0.0.0.0'; I've never heard of using that as a host name for the local host. And, yet, including it in the check for MaServerLocation>>#isLocalHost is a huge performance hit.
Ditch it.

by Chris Muller (ma.chris.m@gmail.com) at April 25, 2018 02:46 AM

empty log message

by Edgar J. De Cleene (edgardec2001@yahoo.com.ar) at April 24, 2018 02:02 PM

- Use of deprecated method fix.
- Include WideString in the #minimumClasses to support special characters.
- Fix for MaClassIdManager>>#initializeMinimumClasses.

by Chris Muller (ma.chris.m@gmail.com) at April 24, 2018 01:22 AM

Fix use of deprecated method.

by Chris Muller (ma.chris.m@gmail.com) at April 24, 2018 01:08 AM