Planet Squeak Code

blogs about Squeak, Pharo, Croquet and family
planet squeak - planet squeak es - planet squeak jp - planet croquet - planet squeak code - planet smalltalk

May 28, 2020

source.squeak.org

FFI-Pools-mt.16

Moves interpretation of <ffiExternalSharedPool> pragma to the definition class where all other pragmas are interpreted and documented already.

Adds simple way to explore pool definitions by calling the method that holds the definition pragmas:

MyExternalPool class >> #win32ExamplePool
<ffiExternalSharedPool>
<ffiPlatformName: 'Win32'>
^ self poolDefinition

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 28, 2020 07:20 AM

May 27, 2020

source.squeak.org

FFI-PoolsTests-mt.8

Fixes typo "Genreator" -> "Generator".

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 03:26 PM

FFI-PoolsTests-monty.2

updated for FFISharedPool renaming

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:52 PM

FFI-Pools-monty.9

renamed every FFISharedPool* class to FFIExternalSharedPool to be more descriptive and avoid conflicting with OSSubProcess. Class comments, generated code prefixes, and the program directory name were also modified for consistency

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:52 PM

FFI-PoolsTests-monty.1

some FFISharedPool test classes

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:48 PM

FFI-Pools-monty.6

use #compilerClass for Squeak compatibility

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:48 PM

Morphic-mt.1662

Adds "recover method versions" to Extras menu. Complements Tools-mt.970.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:30 PM

Tools-mt.970

Adds a simple way to scan the .changes file for verions of methods that are not currently installed. So, you can recover such deletions even if you forgot about or deleted the corresponding change set.

(Note that the code base in ChangeList is scary. I think I added this new feature in a way that matches the existing ones. But we should really clean up here.)

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 02:27 PM

FFI-Kernel-mt.77

Since ExternalAddress makes extensive use of #wordSize, avoid spreading FFIPlatformDescription but re-introduce the message #wordSize instead.

...I hope this has no implications on performance in general, because wordSize was a class-side instVar in ExternalAddress before.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 01:01 PM

FFI-Pools-mt.15

Complements FFI-Kernel-mt.76.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 09:43 AM

FFI-Kernel-mt.76

At least from within FFI package, use FFIPlatformDescription, not the Smalltalk global.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 09:42 AM

FFI-PoolsTests-mt.7

Complements FFI-Kernel-mt.75. Fixes typo "Genreator" -> "Generator".

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 09:12 AM

FFI-Pools-mt.14

Complements FFI-Kernel-mt.75.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 09:11 AM

FFI-Kernel-mt.75

Merges start-up logic from ExternalAddress, ExternalObject, and FFIExternalSharedPool (in FFI-Pools) into a single FFIPlatformDescription (which was FFIExternalSharedPoolPlatform from FFI-Pools).

The list of observers is hard-coded in FFIPlatformDescription class >> #startUp: for now.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 27, 2020 09:09 AM

System-mt.1162

Adds missing #unload.

Follow the discussion here: http://forum.world.st/The-Inbox-System-mt-1161-mcz-td5117249.html

Current issues:
- Maybe rename to HostPlatform
- Maybe (not) model platforms as subclasses of HostPlatform

May 27, 2020 07:40 AM

Tests-ul.434

- speed up DecompilerTests >> #classNames by only computing #category once per class

by Levente Uzonyi (leves@caesar.elte.hu) at May 27, 2020 01:33 AM

May 26, 2020

source.squeak.org

GDB-bgs.3

Execution on non-simulated hardware.

by Boris G. Shingarov (boris@shingarov.com) at May 26, 2020 09:24 PM

VMMaker.gdb-bgs.2747

Execution on non-simulated hardware.
Corresponds to GDB.bgs-2.

by Boris G. Shingarov (boris@shingarov.com) at May 26, 2020 09:20 PM

squeaksource.com

OSProcess-dtl.119

OSProcess 4.6.20

Strings are now immutable in Squeak, fix a test that was writing C null terminators to strings.

Add OSProcess class>>editOutputOf: convenience method for running a command pipeline and returning the output in an editor.

by Dave Lewis (lewis@mail.msen.com) at May 26, 2020 05:55 PM

OSProcess-Base-dtl.72

OSProcess 4.6.20

Strings are now immutable in Squeak, fix a test that was writing C null terminators to strings.

Add OSProcess class>>editOutputOf: convenience method for running a command pipeline and returning the output in an editor.

by Dave Lewis (lewis@mail.msen.com) at May 26, 2020 05:54 PM

OSProcess-Tests-dtl.19

OSProcess 4.6.20

Strings are now immutable in Squeak, fix a test that was writing C null terminators to strings.

Add OSProcess class>>editOutputOf: convenience method for running a command pipeline and returning the output in an editor.

by Dave Lewis (lewis@mail.msen.com) at May 26, 2020 05:54 PM

source.squeak.org

System-mt.1161

Propose SystemPlatform and mechanism to detect platform changes at image startup.

Implementation is based on Monty's FFIExternalSharedPoolPlatform from the FFI-Pools package.

WHY? Can be used to manage platform-specific caches such as FFI's external structures or specifics in other VM plugins.

May 26, 2020 03:28 PM

May 23, 2020

squeaksource.com

NITDMSReader-nice.10

Introduce bulk readValuesForAllsChannels in the easy case.

(NITDMSReader new filename: '/my_directory/my_file.tdms'; readValuesForAllsChannels) inspect.

Easy case is when all channels share same data type and size.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 23, 2020 07:44 AM

May 22, 2020

squeaksource.com

BarnsleyFern-kfr.21

keyboard handling (still bugs) recategorization

May 22, 2020 02:02 PM

BarnsleyFern-kfr.20

Color adjusting. Changed from Object to Morph. moveForward and rotate:

May 22, 2020 01:37 PM

May 21, 2020

squeaksource.com

Doc-tty.4

Added a heuristic for TerseGuide to add appropriate content besides the always appropos 'foo'.

Transcript clear; show:OrderedCollection help

will have the DocletTerseGuide return OrderedCollection and SortedCollection information
f

by tty (gettimothy@zoho.com) at May 21, 2020 07:57 PM

BarnsleyFern-kfr.19

raycast rotation

May 21, 2020 06:55 PM

BarnsleyFern-kfr.18

Raycast first commit

May 21, 2020 05:44 PM

source.squeak.org

Sound-kks.70

Remove redundant override

May 21, 2020 04:14 PM

May 20, 2020

source.squeak.org

FFI-Kernel-mt.74

Adds some support for calling #free from within the image to then also get usable print strings for the instances of ExternalStructure.

Not sure whether it makes sense, but maybe we could store an additional flag in ExternalStructure. Just checking handle for nil seems weird...

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 20, 2020 05:58 PM

FFI-Kernel-mt.73

Refine the print strings for external structure to also indicate pointer types for custom structs.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 20, 2020 05:09 PM

FFI-Tools-mt.5

Better support for NULL pointers in both Inspector and ObjectExplorer.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 20, 2020 04:32 PM

FFI-Kernel-mt.72

Show in a structure's print-string whether it points to NULL.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 20, 2020 04:29 PM

May 14, 2020

source.squeak.org

VMMaker.oscog-eem.2755

Better document callPrimitiveBytecode.

by Eliot Miranda (eliot.miranda@gmail.com) at May 14, 2020 08:54 PM

Kernel-eem.1329

Better document InstructionPrinter>>callPrimitive:

by Eliot Miranda (eliot.miranda@gmail.com) at May 14, 2020 08:49 PM

Compiler-eem.433

Better document SistaV1 bytecode 248

by Eliot Miranda (eliot.miranda@gmail.com) at May 14, 2020 08:48 PM

BytecodeSets.spur-eem.80

Better document InstructionPrinter>>callInlinePrimitive:

by Eliot Miranda (eliot.miranda@gmail.com) at May 14, 2020 08:47 PM

May 13, 2020

source.squeak.org

MorphicTests-nice.63

Let one more MorphicTests pass when ScrollPane useRetractableScrollBars.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 13, 2020 09:34 PM

squeaksource.com

Juggle-ebi.6

empty log message

by Eberhard Gerster (eberhardgerster@t-online.de) at May 13, 2020 03:46 PM

May 12, 2020

source.squeak.org

MorphicTests-nice.62

Let MorphicTests pass when ScrollPane useRetractableScrollBars.

Yeah! And I also put'em where they should, on the left ;)

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 12, 2020 10:26 PM

Tests-nice.433

Acknowledge a reduction of dependencies.

Waouh, good work!

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 12, 2020 09:59 PM

Tools-nice.968

Make explore pointers a bit more robust to broken objects

I have testIntegrityOfDictionaries failing with a Dictionary with tally==nil and array==nil and I would like to chase/inspect/explore pointers...

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 12, 2020 09:49 PM

System-nice.1157

Make inspect pointers a bit more robust to broken objects

I have testIntegrityOfDictionaries failing with a Dictionary with tally==nil and array==nil and I would like to chase/inspect/explore pointers...

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 12, 2020 09:48 PM

May 11, 2020

source.squeak.org

System-tpr.1156

Correct a couple of underscore as assign issues

by tim Rowledge (tim@rowledge.org) at May 11, 2020 01:24 AM

Nebraska-tpr.57

Correct a case of assigning to a block arg

by tim Rowledge (tim@rowledge.org) at May 11, 2020 01:23 AM

EToys-tpr.393

Correct some abuses ofunderscore as assign

by tim Rowledge (tim@rowledge.org) at May 11, 2020 01:21 AM

May 10, 2020

source.squeak.org

Collections-nice.895

Remove byteSize, because super just works
It could have been defined as:
^self size * self bytesPerElement

But super is less dependent on correct redefinition of above two messages:
^ self basicSize * self bytesPerBasicElement

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 07:51 PM

Balloon-nice.33

Move PointArray one level higher.
It is usefull if we want to deprecate IntegerArray...
Also, it shall not reply itself to asIntegerArray.

Provide a correct bytesPerElement taking the two slots occupied by each element into account.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 07:46 PM

Kernel-nice.1328

Return to a safer ClassBuilder.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:38 PM

Sound-nice.69

Use Float32Array explicitely

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:23 PM

Tools-nice.967

asFloat3Array and asFloat64Array

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:22 PM

Graphics-nice.431

Use Float32Array explicitely

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:20 PM

EToys-nice.392

Use Float32Array explicitely

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:20 PM

CollectionsTests-nice.339

Use Float32Array explicitely

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:19 PM

Kernel-nice.1327

Be unsafe for enabling migration of ByteArray

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:14 PM

Collections-nice.894

Clean-up required after the FloatArray rename in preamble and the ByteArray voyage in hierarchy in postcript.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 04:05 PM

squeaksource.com

NITDMSReader-nice.9

Move NITDMSTimestampArray under the brand new RawBitsArray provided by Squeak trunk (6.0).

Also use the new Squeak trunk subclasses of RawBitsArray as recipients.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 03:58 PM

source.squeak.org

Collections-nice.891

Experimental - For discussion

Group some (most) non-pointers collections under an abstract FixedBitWifthArray.
I know, the name is hard to pronounce and thus ugly: it's opened to discussion.

This enables factorization of some methods, for example the trick for atAllPut:
Also notice that most methods are shared between FloatArray and Float64Array.

Introduce SignedByteArray and Long64Array.
These are the only two missing in the panoply.
The heterogeneity of names generally sucks, but we won't change the ByteArray nor WordArray, so probably neither the DoubleByte and DoubleWord Arrays, it follows a legacy-compatible logic.
But for the 32 and 16 bits signed versions, we already have IntegerArray and ShortIntegerArray... What a mess.

We could use better names in the abstract levels too:
For example SignedIntegerArray instead of FixedBitWidthSignedIntegerArray.
Also we could let FloatArray be the abstract class rather than FixedBitWidthFloatArray.
It would be a factory for Float32Array so as to preserve backward compatiblity.

Note that Balloon ShortIntegerArray is currently word rather than doubleByte and should be mutated and moved into this collection. However, it's not easy to mutate because it has both a subclass and instances. We can handle that surgery later.

We can neither easily move ByteArray thru the hierarchy: it is protected because of the fear to break CompiledMethod. Another surgical operation, outside of this proof of concept.

One thing I don't know is the cost of extending intermediate hierachical levels with respect to lookup of message selector. Is it a problem?

While at it, document the internal format of ColorArray.
I hope it's consistent across VMs/platforms/endianness... If it's not, the comment SHALL tell so, that's the least I expect of serious software. I don't know why squeakers are so spare of comments...
Since those classes are good candidates for exchanging data with foreign functions, it's vital to have a minimum of documentation.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 03:58 PM

Collections-nice.893

Create an intermediate abstract level under ArrayedCollection named RawBitsArray for grouping non pointers Arrays and enabling better factorization of some features.

Rename FloatArray -> Float32Array (see preamble of this package) and let FloatArray be the abstract superclass of both Float32Array and Float64Array.

Note: let FloatArray be a factory for creating Float32Array for preserving backward compatibility.

Create asFloat32Array and let asFloatArray default to that.

Move some pre-existing ByteArray, DoubleByteArray, WordArray DoubleWordArray under this hierarchy. For ByteArray, this require some surgery (see postscript of this package).

Create similar SignedByteArray, SignedDoubleByteArray, SignedWordArray, SignedDoubleWordArray for signed integers.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 03:57 PM

squeaksource.com

BulkDataTransfer-nice.2

Provide missing basicReplaceFrom:to:with:startingAt:

Requires the RawBitsArray class from recent Squeak trunk (6.0).

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 03:56 PM

Compiler.V3-dtl.432

Name: Compiler-nice.432
Author: nice
Time: 10 May 2020, 12:50:37.677856 pm
UUID: f6faf998-9905-4fbd-9bc4-66a2e9f8bc93
Ancestors: Compiler-nice.431

Fix Decompiler after correction byteCodes generated by inlined #caseOf: and recompile all senders of caseOf: in postscript.

Note: I have changed the logic a little bit:
- the ancient CaseFlag is replaced by OtherwiseFlag (that's the purpose, we are trying to detect last case before otherwise:).
- CascadeFlag is replaced by CaseFlag as soon as we have detected a potential caseOf:.

I never put so many Halt in code before having it right. Good luck to the next one wanting to change the Decompiler...

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:41 PM

Compiler.V3-dtl.431

Name: Compiler-nice.431
Author: nice
Time: 10 May 2020, 9:46:13.694981 am
UUID: d25683a4-9332-443c-9c2f-6d331d7993a6
Ancestors: Compiler-nice.430, Compiler-ct.425

Merge Compiler-ct.425, fix the AST produced by Decompiler so that it can generate correct byteCodes.

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:39 PM

Compiler.V3-dtl.430

Name: Compiler-nice.430
Author: nice
Time: 10 May 2020, 9:27:14.242821 am
UUID: 7ba9b09a-1ebd-4181-bf85-2dc912f42c9d
Ancestors: Compiler-nice.429

Fixup: caseError must be sent to the receiver of the caseOf: message, not to the receiver of the home method.

For this, we must also dup/pop in the last case whenever there is no otherwise argument.

See http://forum.world.st/Problems-with-caseError-tp5111930p5112255.html

TODO: also fix the Decompiler...

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:31 PM

Compiler.V3-dtl.429

Name: Compiler-nice.429
Author: nice
Time: 9 May 2020, 12:34:59.991171 am
UUID: 3f35c872-f6c4-45b8-9d7f-785b8f9f70f8
Ancestors: Compiler-nice.428, Compiler-ct.422

Merge Compiler-ct.422 which allow to cancel compilation if an unused variable is detected

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:30 PM

Compiler.V3-dtl.428

Name: Compiler-nice.428
Author: nice
Time: 9 May 2020, 12:24:17.077079 am
UUID: 04b649b2-1b1b-486b-aa82-15b219803431
Ancestors: Compiler-nice.427

Use the idea from Compiler-ct.423: define selectFrom:to:during: in Parser for handling temporary change of text selection - see method comment.

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:28 PM

Compiler.V3-dtl.427

Includes Compiler-ct.424, Compiler-ct.426, and Compiler-nice.427

Name: Compiler-nice.427
Author: nice
Time: 8 May 2020, 6:59:54.689381 pm
UUID: e2c7d4b2-b79d-4d02-a576-872003ec88ed
Ancestors: Compiler-ct.426

Avoid sending select/deselect to the editor in case of interactive correction.

These are st80 specific, forcing Morphic to have stubs, and are considered too intrusive: we should better let such responsibility to the editor, that's its own business.

The send of #select was un-necessary anyway, because selectFrom:to: will send #selectAndScroll, which will send #select.

However, deselect was still required for st80, so replace the sequence deselect; selectInvisiblyFrom:to: with a single message #selectIntervalInvisibly: which is provided by ST80-nice.254 and Morphic-nice.1657.

Why is this needed? As explained in the st80 commit, we want to restore the (eventually slightly modified) user selection once the corrections are performed.

We must do that at each correction, so as to correctly track addition or delection of characters (the selection interval might need to grow or shrink accordingly).

But we do not want to make the user selection visible instantly, otherwise the selection will go back to user selection and forth to next zone of interactive correction repeatedly creating an annoying flashing effect.

Care to let future reader know about it with a comment, there's nothing that obvious which could make the comment superfluous!

Name: Compiler-ct.426
Author: ct
Time: 28 March 2020, 1:22:49.51256 am
UUID: 761924e9-77a0-1144-9533-8a7c22c03332
Ancestors: Compiler-ct.424

Fixes a bug regarding decompilation of special selectors.

The following did not work before:

(OrderedCollection >> #asArray) decompile generate valueWithReceiver: {42} asOrderedCollection arguments: #(). "MessageNotUnderstood: OrderedCollection>>#Array=>Array"

See also this thread: http://forum.world.st/The-Inbox-EToys-ct-367-mcz-tp5105507p5114020.html

Name: Compiler-ct.424
Author: ct
Time: 26 March 2020, 5:19:54.71456 pm
UUID: 10f4405b-896c-3149-b76b-b9cd45ca5734
Ancestors: Compiler-nice.420

Fixes a bug regarding decompilation of FullBlockClosures

Sample to reproduce: [^ true] decompile should NOT return {[]}!

by Dave Lewis (lewis@mail.msen.com) at May 10, 2020 03:27 PM

source.squeak.org

Kernel-nice.1326

Fix testMoreThanOnceForEqualBlocks by using an IdentityDictionary in #once.

Note that the postscript does convert the existing caches into IdentityDictionary.
This is necessary in order to make the test pass if it was ran once previously.

Refactor caseError so that the error message can be internationalized (change from ct, to be finished by sending translated if we wish).

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 01:26 PM

Collections-nice.892

Fixup: of course, we need a regex if we want to use the negative lookahead.
Also $: needs to be escaped.

Apologies for the quality of previous commit.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 12:41 PM

Compiler-nice.432

Fix Decompiler after correction byteCodes generated by inlined #caseOf: and recompile all senders of caseOf: in postscript.

Note: I have changed the logic a little bit:
- the ancient CaseFlag is replaced by OtherwiseFlag (that's the purpose, we are trying to detect last case before otherwise:).
- CascadeFlag is replaced by CaseFlag as soon as we have detected a potential caseOf:.

I never put so many Halt in code before having it right. Good luck to the next one wanting to change the Decompiler...

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 10:50 AM

Compiler-nice.431

Merge Compiler-ct.425, fix the AST produced by Decompiler so that it can generate correct byteCodes.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 10, 2020 07:46 AM

May 09, 2020

squeaksource.com

NITDMSReader-nice.8

Even if there is a new list of objects, we must still try to mergeIndexFrom: previous list.
Indeed, it happens that object in the new list declare having identical index than same object on previous list...

Also the numberOfRawDataChunks must be queried only using updated object list, otherwise attempts to obtain totalRawDataSizeInBytes from a NITDMSRawDataIndexIdentical will fail.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 09, 2020 08:57 PM

Juggle-ebi.5

empty log message

by Eberhard Gerster (eberhardgerster@t-online.de) at May 09, 2020 03:07 PM

May 06, 2020

squeaksource.com

SBE-Quinto-jec.1

First checkin

by Jon Clark (n6nq@roadrunner.com) at May 06, 2020 10:41 PM

May 05, 2020

source.squeak.org

EToys-ct.362

Complements Tools-ct.900 (Inspector refactoring)

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 05, 2020 03:47 PM

Protocols-ct.68

Complements Tools-ct.900 (Inspector refactoring)

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 05, 2020 03:46 PM

Tools-ct.906

I mixed up #asTextFromHtml with the *widgets version, sorry :(

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 05, 2020 03:45 PM

Tools-ct.907

Fixes a styling bug in ContextVariableInspectors, we don't need fieldList, we have slotSpecs.

For example, if you debug the following example before loading this commit, x will be styled wrong:

42 ifNotNil: [:x | x asWords]

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 05, 2020 03:43 PM

System-ct.1156

Fixes double prompt for handling a dropped file when the first dialog was cancelled. Cleans up duplication between ExternalDropHandler >> #lookupServiceBasedHandler: and DefaultExternalDropHandler by deprecating the first one. Maybe we can also remove this method?

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at May 05, 2020 03:41 PM

squeaksource.com

EbisWindow-ebi.13

postscript added

by Eberhard Gerster (eberhardgerster@t-online.de) at May 05, 2020 02:10 PM

EbisWindow-ebi.12

Extensions

by Eberhard Gerster (eberhardgerster@t-online.de) at May 05, 2020 01:59 PM

EbisWindow-ebi.11

Extensions berichtigt

by Eberhard Gerster (eberhardgerster@t-online.de) at May 05, 2020 01:36 PM

source.squeak.org

Morphic-ct.1654

Deprecates #lockUnlockMorph in favor of #toggleLocked

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at May 05, 2020 01:35 PM

May 04, 2020

source.squeak.org

Chronology-Core-cmm.56

Building on Chronology-Core-ul.54:

- Let #microsecondClockValue complement #millisecondClockValue, to offer higher precision.
- New constructor, DateAndTime class>>#utcMicroseconds:, complements the above.
- Make Time class>>#utcMicrosecondClock private to help avoid accidental improper use with the above.

by Chris Muller (ma.chris.m@gmail.com) at May 04, 2020 11:42 PM

squeaksource.com

NITDMSReader-nice.7

Simplify allocation: don't mess with the arrayClass own business, it knows how.
Add a few class comments.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 06:41 PM

ConfigurationOfOCompletion-mt.77

Adds version 1.4.5 for Squeak 5.3 and 6.0alpha.

May 04, 2020 04:37 PM

source.squeak.org

Morphic-mt.1634

Fixes bug in accepting dialog requests whose input does not appear changed due to text undo. See http://forum.world.st/FillInTheBlankMorph-gets-amorphous-when-cancelling-edits-with-ALT-Z-td5115936.html

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 04, 2020 02:07 PM

Morphic-mt.1465

Fixes bug in accepting dialog requests whose input does not appear changed due to text undo. See http://forum.world.st/FillInTheBlankMorph-gets-amorphous-when-cancelling-edits-with-ALT-Z-td5115936.html

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 04, 2020 02:05 PM

Morphic-mt.1297

Fixes bug in accepting dialog requests whose input does not appear changed due to text undo. See http://forum.world.st/FillInTheBlankMorph-gets-amorphous-when-cancelling-edits-with-ALT-Z-td5115936.html

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 04, 2020 02:03 PM

Tools-jr.966

Avoid inlining of #class for proxies/decorators in Inspector.

The instance variables and indexed variables are accessed by sending a message to the inspected object, which probably passes through any transparent proxies. If the list of variables is just obtained from "object class" (which will not actually send a message to the object), the list will contain the variables of the proxy. Thus, there is a mismatch between the inspector's schema of the object (the fields) and the schema which is queried for values (the schema of the object behind the proxy).

Also send a message to find out the class, so it is aligned with how the values are retrieved.

Also mind this when initializing the styler.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 04, 2020 01:57 PM

squeaksource.com

NITDMSReader-nice.6

Fixup inst. var. refactoring typo s/dimensio/dimension/
Remove noteNoRawData (same as super)

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 01:32 PM

source.squeak.org

Morphic-mt.1655

In fill-in dialogs, it should always be possible to accept the contents even if they are empty or appear otherwise unchanged.

Note that "hasUnacceptedEdits: true" might not be needed anymore. Anyway, the MorphicToolBuilder gave me some hints on how #askBeforeDiscardingEdits and #alwaysAccept are related.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at May 04, 2020 09:28 AM

squeaksource.com

NITDMSReader-nice.5

Fixup: better merging the previous metaData before counting the number of values.

Thus, the lastNumberOfValues is gone from channel object.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 01:59 AM

NITDMSReader-nice.4

Fixup the count of values for channels: the number of consecutive identical chunks was not taken into account!

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 01:38 AM

NITDMSReader-nice.3

Many changes, including:
- Use OrderedDictionary so as to preserve order of the list of objects in metaData, but also of groups and channels.
- Let Channel object have dataTypes and values.
- Introduce allocation of Smalltalk BitsArrays corresponding to foreign dataTypes.
- Implement reading a single channel as a POC. This uses the BulkDataTransfer package.
TODO: read multiple channels with same tricks.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 12:25 AM

BulkDataTransfer-nice.1

First version for bulk transfer of raw data using Bit Blit when both srcArray and dstArray class isBits.

Especially useful when reading foreign binary data files.

Beware, this version is not much tested yet.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 04, 2020 12:15 AM

May 03, 2020

squeaksource.com

NITDMSReader-nice.2

Introduce NITDMSTimestampArray for representing arrays of accurate National Instrument 128 bits TDMS Timestamp.

See https://www.ni.com/fr-fr/support/documentation/supplemental/08/labview-timestamp-overview.html

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 03, 2020 06:59 PM

source.squeak.org

Chronology-Core-ul.54

Automatically update the time zone cached by the VM every 30 minutes:
- copied and modified Eliot's accessor to primitive 243 to do the actual updating (Time class >> #primitiveUpdateTimeZone)
- update happens when Time class >> #posixMicrosecondClockWithOffset or Time class >> #posixMicrosecondClockWithOffset: are sent. Other low-level accessors of the VM time zone are not updating the cached time zone value (e.g. Time class >> #localMicrosecondClockPrimitive)
- actual update happens in Time class >> #updateTimeZoneCacheAt:
- postscript activates the update mechanism which can be turned off by evaluating [ Time classPool at: #UpdateVMTimeZoneCacheAt put: nil ]

Other changes:
- added Time class >> #posixUtcMicrosecondClock to have a fast way to access that raw timestamp
- added MicrosecondsBetweenPosixEpochAndSqueakEpoch to ChronologyConstants to support the above
- do not send Time class >> #initialize from DateAndTime class >> #startUp:. Let Time have its own #startUp: method instead. Postscript adds Time to the startUpList before DateAndTime.
- do not send super initialize from DateAndTime class >> #initialize
- do not send #automaticTimezone to DateAndTime on startup, because that's just an accessor with no side effects
- do not overwrite set variables in Time class >> #initialize
- assert that the value passed to Time class >> #clockPolicy: is one of the known values

by Levente Uzonyi (leves@caesar.elte.hu) at May 03, 2020 06:45 PM

May 02, 2020

squeaksource.com

CommandShell-dtl.109

CommandShell 4.7.10
Additional updates for Cuis support. FileMan updates and fix tests to match.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 10:00 PM

CommandShell-Commands-dtl.25

CommandShell 4.7.10
Additional updates for Cuis support. FileMan updates and fix tests to match.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 10:00 PM

CommandShell-Base-dtl.72

CommandShell 4.7.10
Additional updates for Cuis support. FileMan updates and fix tests to match.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 09:59 PM

CommandShell-Piping-dtl.29

CommandShell 4.7.10
Additional updates for Cuis support. FileMan updates and fix tests to match.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 09:59 PM

CommandShell-Tests-dtl.20

CommandShell 4.7.10
Additional updates for Cuis support. FileMan updates and fix tests to match.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 09:59 PM

source.squeak.org

Collections-nice.891

Experimental - For discussion

Group some (most) non-pointers collections under an abstract FixedBitWifthArray.
I know, the name is hard to pronounce and thus ugly: it's opened to discussion.

This enables factorization of some methods, for example the trick for atAllPut:
Also notice that most methods are shared between FloatArray and Float64Array.

Introduce SignedByteArray and Long64Array.
These are the only two missing in the panoply.
The heterogeneity of names generally sucks, but we won't change the ByteArray nor WordArray, so probably neither the DoubleByte and DoubleWord Arrays, it follows a legacy-compatible logic.
But for the 32 and 16 bits signed versions, we already have IntegerArray and ShortIntegerArray... What a mess.

We could use better names in the abstract levels too:
For example SignedIntegerArray instead of FixedBitWidthSignedIntegerArray.
Also we could let FloatArray be the abstract class rather than FixedBitWidthFloatArray.
It would be a factory for Float32Array so as to preserve backward compatiblity.

Note that Balloon ShortIntegerArray is currently word rather than doubleByte and should be mutated and moved into this collection. However, it's not easy to mutate because it has both a subclass and instances. We can handle that surgery later.

We can neither easily move ByteArray thru the hierarchy: it is protected because of the fear to break CompiledMethod. Another surgical operation, outside of this proof of concept.

One thing I don't know is the cost of extending intermediate hierachical levels with respect to lookup of message selector. Is it a problem?

While at it, document the internal format of ColorArray.
I hope it's consistent across VMs/platforms/endianness... If it's not, the comment SHALL tell so, that's the least I expect of serious software. I don't know why squeakers are so spare of comments...
Since those classes are good candidates for exchanging data with foreign functions, it's vital to have a minimum of documentation.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 02, 2020 05:41 PM

squeaksource.com

CommandShell-dtl.108

CommandShell 4.7.9
Remove use of #release. This was used for removing dependencies, but appears to be unnecessary as there are no cyclic dependencies. Cuis does not implement #release, so get rid of it now.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 04:03 PM

CommandShell-Morphic-dtl.4

CommandShell 4.7.9
Remove use of #release. This was used for removing dependencies, but appears to be unnecessary as there are no cyclic dependencies. Cuis does not implement #release, so get rid of it now.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 04:03 PM

CommandShell-Piping-dtl.28

CommandShell 4.7.9
Remove use of #release. This was used for removing dependencies, but appears to be unnecessary as there are no cyclic dependencies. Cuis does not implement #release, so get rid of it now.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 04:02 PM

CommandShell-UI-dtl.22

CommandShell 4.7.9
Remove use of #release. This was used for removing dependencies, but appears to be unnecessary as there are no cyclic dependencies. Cuis does not implement #release, so get rid of it now.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 04:02 PM

CommandShell-Base-dtl.71

CommandShell 4.7.9
Remove use of #release. This was used for removing dependencies, but appears to be unnecessary as there are no cyclic dependencies. Cuis does not implement #release, so get rid of it now.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 04:01 PM

CommandShell-dtl.107

CommandShell 4.7.8
Additional updates for Cuis. Make file redirection work. Eliminate unneeded calls to release. Update fileExists for FileMan. Document TtyTextMorphForEditView, no longer needed for recent Squeak but required for older Squeak and probably Pharo.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 02:24 PM

CommandShell-Base-dtl.70

CommandShell 4.7.8
Additional updates for Cuis. Make file redirection work. Eliminate unneeded calls to release. Update fileExists for FileMan. Document TtyTextMorphForEditView, no longer needed for recent Squeak but required for older Squeak and probably Pharo.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 02:24 PM

CommandShell-Commands-dtl.24

CommandShell 4.7.8
Additional updates for Cuis. Make file redirection work. Eliminate unneeded calls to release. Update fileExists for FileMan. Document TtyTextMorphForEditView, no longer needed for recent Squeak but required for older Squeak and probably Pharo.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 02:24 PM

CommandShell-Morphic-dtl.3

CommandShell 4.7.8
Additional updates for Cuis. Make file redirection work. Eliminate unneeded calls to release. Update fileExists for FileMan. Document TtyTextMorphForEditView, no longer needed for recent Squeak but required for older Squeak and probably Pharo.

by Dave Lewis (lewis@mail.msen.com) at May 02, 2020 02:23 PM

source.squeak.org

VMMaker.oscog-nice.2750

Extend the capability of SmartSyntaxPluginCodeGenerator to handle DoubleByte and DoubleWord Arrays.

Remove the #debugCode: from the prolog of LargeIntegersPlugin primitives.
This was used only for printing a console trace upon primitive entry.
That's not how we are going to debug the plugin.
We are going to debug either with a VM Simulation, or via a modern debugger like gdb (cough!) if ever the problem lies in code generation itself.

Keep only two sends of debugCode: tracing case of un-normalized inputs.

Note: most of the SmartSyntaxPluginCodeGenerator capabilities/messages are currently unused.
See #initializeCTranslationDictionary.
The lack of examples make it very difficult to follow the code, or even guess the purpose.
No need to say that comments will not help.
Comments? who ever needed a comment in Smalltalk?
With greatest shame, I must confess, I need comments sometimes!
With such property, the code may live very long untouched, put your dirty nose out of this code ;)

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 02, 2020 10:34 AM

CollectionsTests-nice.338

Introduce tests for Float64Array

by David T. Lewis (lewis@mail.msen.com) at May 02, 2020 01:05 AM

Collections-nice.890

Introduce Float64Array.
This will be fast once the Float64ArrayPlugin is delivered.

by David T. Lewis (lewis@mail.msen.com) at May 02, 2020 01:05 AM

May 01, 2020

source.squeak.org

Tools-jr.966

Avoid inlining of #class for proxies/decorators in Inspector.

The instance variables and indexed variables are accessed by sending a message to the inspected object, which probably passes through any transparent proxies. If the list of variables is just obtained from "object class" (which will not actually send a message to the object), the list will contain the variables of the proxy. Thus, there is a mismatch between the inspector's schema of the object (the fields) and the schema which is queried for values (the schema of the object behind the proxy).

Also send a message to find out the class, so it is aligned with how the values are retrieved.

Also mind this when initializing the styler.

May 01, 2020 10:16 PM

VMMaker.oscog-nice.2749

Introduce Float64ArrayPlugin which is same as FloatArrayPlugin but for Float64Array.
This can work in Spur which provides DoubleWordArray.

This requires one VM function which where not in API: isLong64s().
Thus it will require a new platforms/Cross/vm/sqVirtualMachine.[ch] definition of interpreter proxy.
While at it, also add isShorts() to the API.
Hence increment the VM_PROXY_MINOR.

Introduce a new primitiveFromFloat64Array to FloatArrayPlugin to initialize a FloatArray from a Float64Array.
This is required for example for Smallapack when coercing single<->double precision.
If this plugin is regenerated, then platforms/Cross/vm/sqVirtualMachine.[ch] must be upgraded accordingly.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at May 01, 2020 09:22 PM

Kernel-eem.1325

Minor clean-up from the last commit. We don't really need compiledBlock[:], so do without.

by Eliot Miranda (eliot.miranda@gmail.com) at May 01, 2020 05:38 PM

Kernel-eem.1324

Rename BlockClosure's startpc inst var to startpcOrMethod, to reduce confusion in FullBlockClosure.

In FullBlockClosure replace sends of self compiledBlock with startpcOrMethod. Simplify FullBlockClosure>>#=.

by Eliot Miranda (eliot.miranda@gmail.com) at May 01, 2020 05:26 PM

Monticello-eem.717

Sort repositories in the Monticello browser, useful if, as in a VMMaker image with Squeak trunk github support included, one has nearly 40 repositories.

by Eliot Miranda (eliot.miranda@gmail.com) at May 01, 2020 05:20 PM

squeaksource.com

NITDMSReader-nice.1

NITDMSReader is a package for reading files abiding to National Instrument TDMS format 1.0 and 2.0.

This file format is used for acquisition of time series with the family of National Instrument softwares like LabView (TM) & al

The format is described here:
https://www.ni.com/fr-fr/support/documentation/supplemental/07/tdms-file-format-internal-structure.html

This first version does not yet read the contents of rawData.
It just scans the metaData index, and answer a NITDMSFile object reflecting the hierarchical structure of the file: file / group / channel

Use it with:

(NITDMSReader new filename: '/my_directory/my_file.tdms'; scan) explore

Note: National Instrument also provides an API and distribute binary DLL for reading such files on Windows and Macos which I already interfaced in VW.
But there are some restrictions in usage and redistribution. Also the binaries are quite huge compared to what is really needed for a quick start, and distributing binaries is always a pain for packaging reasons.
In Squeak, I'm confident that we can optimize reading enough by using BitBlt tricks for data transfer.

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at May 01, 2020 01:47 PM

April 24, 2020

source.squeak.org

Morphic-ct.1650

Proposal: Fix a small glitch in menu selection logic. In a labeled menu, you can hover the first item and then hover the title/close button, but the selection will be only reset when the whole menu is unhovered. This is confusing because pressing the mouse button does not actually trigger the menu item.

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at April 24, 2020 04:40 PM

VMMaker.oscog-eem.2744

Slang: Make sure plugins include sqMathShim.h immediately aftyer math.h so that BIT_IDENTICAL_FLOATING_POINT works for plugins too.

by Eliot Miranda (eliot.miranda@gmail.com) at April 24, 2020 04:42 AM

April 23, 2020

source.squeak.org

VMMaker.oscog-eem.2743

BIT_IDENTICAL_FLOATING_POINT: Add a feature flag to vmParameterAt: 65 to show that the Vm was compiled with BIT_IDENTICAL_FLOATING_POINT.

Revert the exclusion of the machine code square root funcitons if BIT_IDENTICAL_FLOATING_POINT is defined. sqrt is rounded consistently across platforms.

by Eliot Miranda (eliot.miranda@gmail.com) at April 23, 2020 07:52 PM

squeaksource.com

Collections.V3-dtl.888

Name: Collections-mt.888
Author: mt
Time: 20 April 2020, 10:16:10.896894 am
UUID: ecc786d1-496b-0e4d-8f51-3abcff1823db
Ancestors: Collections-mt.887

Fixes mistake in previous commit. Sorry, it's Monday. --- Complements CollectionsTests-mt.337, which found the mistake.

by Dave Lewis (lewis@mail.msen.com) at April 23, 2020 12:45 AM

April 22, 2020

source.squeak.org

Tools-ct.963

Fixes a bug in FileList2 >> #endingSpecs, caused by introduction of read-only literals (see Kernel-eem.1318). Fixes MorphicUIManagerTest >> #testShowAllBinParts. Also contains some refactoring of related stuff.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 22, 2020 10:29 AM

Tools-mt.964

Merges Christoph's fix for FileList2 concerning the ModificationForbidden "error".

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 22, 2020 10:28 AM

April 20, 2020

source.squeak.org

Collections-mt.887

Let <img> tags honor Squeak's "High DPI" setting.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 20, 2020 07:41 AM

April 19, 2020

source.squeak.org

VMMaker-dtl.415

VMMaker 4.19.1
Required for Squeak trunk Collections-eem.885 and above.

Install primitiveArrayBecomeOneWayNoCopyHash as primitive 248, replacing obsolete primitiveInvokeObjectAsMethod.

Fix 128 primitiveArrayBecome to match oscog logic. This a fix from VMMaker.oscog-eem.647 which Eliot explained as follows:
Fix primitiveArrayBecome (the two-way become); it should /not/
specify copyHash. ObjectMemory ignores the copyHash flag when doing a two-way become, hence the wrong sense of the flag had
no effect. Hence correctly evaluate testBecomeIdentityHash.

The primitives now do this:
primitive 72: twoWay: false copyHash: true
primitive 128: twoWay: true copyHash: false
primitive 248: twoWay: false copyHash: false
primitive 249: twoWay: false copyHash: <arg 3 value>

by David T. Lewis (lewis@mail.msen.com) at April 19, 2020 09:30 PM

squeaksource.com

System.V3-dtl.1154

Name: System-eem.1154
Author: eem
Time: 19 January 586155, 1:08:16.564616 am
UUID: 774a2409-d212-1a47-8988-0250ab5f13d2
Ancestors: System-mt.1153

Spelling in a class comment

by Dave Lewis (lewis@mail.msen.com) at April 19, 2020 08:53 PM

System.V3-dtl.1153

Name: System-mt.1153
Author: mt
Time: 17 April 2020, 5:16:44.35686 pm
UUID: ed1deacd-13b8-9144-850a-a91589b99e87
Ancestors: System-eem.1152

Adds a query for information about the CI status to the system version. Why? Because the system version shapes that query.

by Dave Lewis (lewis@mail.msen.com) at April 19, 2020 08:49 PM

Collections.V3-dtl.886

Name: Collections-mt.886
Author: mt
Time: 17 April 2020, 4:56:30.33186 pm
UUID: c7d64e56-0d06-e34f-8a61-8f5a7eb9277d
Ancestors: Collections-eem.885

To our HTML-to-Text converter, add support for <img> tags. Either download an image (or picture) from the Web or evaluate some code to retrieve either a Form or a Morph. As documented in #httpGetImage:, this complements the support of "code://" in TextURL.

by Dave Lewis (lewis@mail.msen.com) at April 19, 2020 08:42 PM

Collections.V3-dtl.885

NOTE - for interpreter VM, requires VMMaker-dtl.415 for required primitive support. Older VMs will not work (but Cog VM has the needed primitives).

Name: Collections-eem.885
Author: eem
Time: 15 April 2020, 4:37:54.800631 pm
UUID: 45d219d3-6ed0-4401-a820-44eebe21d71a
Ancestors: Collections-eem.883, Collections-dtl.884

Switch elementsForwardIdentityTo: to not copy the hash, see http://forum.world.st/How-to-become-immediate-objects-td5114931.html.
Add elementsForwardIdentityAndHashTo: for the old behavior.

by Dave Lewis (lewis@mail.msen.com) at April 19, 2020 08:40 PM

April 18, 2020

source.squeak.org

System-eem.1154

Spelling in a class comment

by Eliot Miranda (eliot.miranda@gmail.com) at April 18, 2020 07:20 PM

MonticelloConfigurations-dtl.167

When saving an MCConlfiguration, let the comment editor check on accept to ensure that editing is complete before proceeding to save. On cancel, undo changes to the name. Edit window stays open until user confirms, and is closed on save or cancel.

April 18, 2020 06:44 PM

squeaksource.com

ParrotTalk-rww.56

make sure to return frames from #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 18, 2020 05:13 PM

ParrotTalk-rww.56

make sure to return frames from #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 18, 2020 05:13 PM

ThunkStack-rww.29

make sure to return frames from #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 18, 2020 05:13 PM

ThunkStack-rww.29

make sure to return frames from #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 18, 2020 05:13 PM

source.squeak.org

Tests-ct.431

Complements Monticello-ct.717.

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at April 18, 2020 02:46 PM

Monticello-ct.717

Fixes regression in the dependency sorter that was introduced via Monticello-ct.715. Restore the original semantics of #orderedItems, meaning to return all items that were added to the sorter. Rename (and reorder) instance variable to avoid confusion with the changed behavior of #orderedItems. Note that #orderedItems is no quick getter any more. Add some documentation. Note that I did not document the instvars 'required' and 'provided' coz I don't have an appropriate overview of them at the moment.

Thanks to Jakob (jr) for the help! :-)

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at April 18, 2020 02:43 PM

MonticelloConfigurations-dtl.166

Provide a proper editor dialog for the MCM version comment entry as part of the MCConfigurationBrowser store operation.

April 18, 2020 02:20 AM

April 17, 2020

squeaksource.com

ThunkStack-rww.28

use the return frame for #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 17, 2020 09:46 PM

ThunkStack-rww.28

use the return frame for #upcall:/#downcall:

by Robert Withers (robert.withers@protonmail.com) at April 17, 2020 09:46 PM

SSL-rww.59

added SSLFrame

by Robert Withers (robert.withers@protonmail.com) at April 17, 2020 09:44 PM

SSL-rww.59

added SSLFrame

by Robert Withers (robert.withers@protonmail.com) at April 17, 2020 09:44 PM

source.squeak.org

MonticelloConfigurations-mt.160

Shows the current CI status in the update dialog.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 03:17 PM

System-mt.1153

Adds a query for information about the CI status to the system version. Why? Because the system version shapes that query.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 03:16 PM

Morphic-mt.1647

Allow users to copy pieces of text from a dialog window's message.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 02:58 PM

CollectionsTests-mt.336

Complements Collections-mt.886

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 02:57 PM

Collections-mt.886

To our HTML-to-Text converter, add support for <img> tags. Either download an image (or picture) from the Web or evaluate some code to retrieve either a Form or a Morph. As documented in #httpGetImage:, this complements the support of "code://" in TextURL.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 02:56 PM

Morphic-mt.1646

Complement Graphics-mt.430.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 02:53 PM

Graphics-mt.430

Adds a simple way to use forms as text anchors without exposing Morphic, where such anchors are currently implemented.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at April 17, 2020 02:52 PM

Compiler-nice.427

Fixup: caseError must be sent to the receiver of the caseOf: message, not to the receiver of the home method.

For this, we must also dup/pop in the last case whenever there is no otherwise argument.

TODO: also fix the Decompiler...

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at April 17, 2020 10:52 AM

April 15, 2020

squeaksource.com

ThunkStack-rww.27

DuplexThunk naming.

by Robert Withers (robert.withers@protonmail.com) at April 15, 2020 05:53 AM

ThunkStack-rww.27

DuplexThunk naming.

by Robert Withers (robert.withers@protonmail.com) at April 15, 2020 05:53 AM

SSL-rww.55

cleanup. DuplexThunk naming.

by Robert Withers (robert.withers@protonmail.com) at April 15, 2020 05:53 AM

SSL-rww.55

cleanup. DuplexThunk naming.

by Robert Withers (robert.withers@protonmail.com) at April 15, 2020 05:53 AM

source.squeak.org

GraphicsTests-eem.54

Add the more complete set of tests for the JPEGReadWriter2Plugin from VMMaker to the base JPEGReadWriter2Test

by Eliot Miranda (eliot.miranda@gmail.com) at April 15, 2020 12:36 AM

April 14, 2020

squeaksource.com

SSL-rww.54

nil out listener on stop.

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:52 PM

SSL-rww.54

nil out listener on stop.

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:52 PM

ThunkStack-rww.26

check for nil listener

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:52 PM

ThunkStack-rww.26

check for nil listener

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:52 PM

SSL-rww.53

encoder mods

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:47 PM

SSL-rww.53

encoder mods

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:47 PM

ThunkStack-rww.25

renamed to #cipherName

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

ThunkStack-rww.25

renamed to #cipherName

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

SSL-rww.52

renamed #cipherName and fragmentation

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

SSL-rww.52

renamed #cipherName and fragmentation

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

ParrotTalk-rww.55

renamed #cipherName

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

ParrotTalk-rww.55

renamed #cipherName

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 11:28 PM

source.squeak.org

VMMaker.oscog-eem.2741

Spur: Make sure that becomeForward: answers a useful error code for read-only targets (when copyHash is true).

by Eliot Miranda (eliot.miranda@gmail.com) at April 14, 2020 09:03 PM

squeaksource.com

SSL-rww.51

making halfThunks

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 07:56 PM

SSL-rww.51

making halfThunks

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 07:56 PM

CryptographyKeyExchange-rww.15

protected #sharedKeyPadPositiveByteArray in case of a nil sharedKey.

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 07:55 PM

CryptographyKeyExchange-rww.15

protected #sharedKeyPadPositiveByteArray in case of a nil sharedKey.

by Robert Withers (robert.withers@protonmail.com) at April 14, 2020 07:55 PM

April 13, 2020

source.squeak.org

VMMaker.oscog-eem.2740

Simulation UI: match VMMakerUI-eem.26

by Eliot Miranda (eliot.miranda@gmail.com) at April 13, 2020 12:07 AM

VMMakerUI-eem.26

Move handleListenEvent: into the UI package and fix the sharing of the displayForm with displayView image. Now displayForm == displayView image is true after initialization.

by Eliot Miranda (eliot.miranda@gmail.com) at April 13, 2020 12:05 AM

April 12, 2020

source.squeak.org

GDB-bgs.1

Initial import of GDB Remote Client interface

by Boris Shingarov (boris@shingarov.com) at April 12, 2020 11:54 PM

ST80-dtl.250

Do not burn CPU if there is nothing to do. If the ScreenController is the only scheduled controller available, then call interActivityPause before rescheduling it.

by David T. Lewis (lewis@mail.msen.com) at April 12, 2020 07:57 PM

April 06, 2020

squeaksource.com

SBE-Quinto-MklBnk.3

3a prueba

by Mkl Bnk (mkbk.cloud@gmail.com) at April 06, 2020 03:29 PM

Smallapack-Packed-nice.2

Fixup: replace two references to CArrayAccessor with ExternalLongArray

Those references where due to porting from VW.

April 06, 2020 01:44 PM

April 05, 2020

squeaksource.com

Startup-SqueakSource-dtl.8

Give names to all of the SS background processes so they are understandable in the process browser.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.7

All oscog VMs have hard-coded SIGUSR1 handler, change startUp: to check for this.
Remove trace: output from socket leak monitor and image save method, not required in production use.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.6

Disable the forkSqueak version of image save, reverting to the original foreground image save as currently running on squeaksource.com. Need further testing before switching to forkSqueak in production, so leave it out for now.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.5

Do no activate SIGUSR1 handler if OSPP is from oscog branch, which does not include necessary primitive support.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.4

Change the image save process to use forkSqueak to save the image in a background OS process. This allows the foreground squeaksource image to continue running without blocking the VM. Saving is done at low OS process priority for minimal overall system impact.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.3

Add the SocketMonitorProcess to SSImageInit. Previously this was running from a workspace in the squeaksource.com image.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.2

Add the image save process to SSImageInit. Previously this was running from a workspace in the squeaksource.com image.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

Startup-SqueakSource-dtl.1

Initial save of squeaksource.com startup code running on http://www.squeaksource.com as of January 2019.

by Dave Lewis (lewis@mail.msen.com) at April 05, 2020 09:51 PM

March 29, 2020

squeaksource.com

CryptographyHashingTests-ul.3

- changed HashFunctionTest class >> #inputs to return a literal instead of computing the value, so that the code does not depend on the implementation of Random
- removed the unused Inputs class variable

by Levente Uzonyi (leves@elte.hu) at March 29, 2020 03:43 PM

March 28, 2020

squeaksource.com

Collections.V3-dtl.884

Name: Collections-dtl.884
Author: dtl
Time: 28 March 2020, 8:45:29.299371 pm
UUID: c6979179-ecac-4854-a979-c9dd2dcaf88e
Ancestors: Collections-dtl.883, Collections-kfr.878

Merge. Add #clear to Transcript window menu. Proposed by Christoph in Collections-ct.869 (now in treated inbox) and merged from Karl's update in Collections-kfr.878.

by Dave Lewis (lewis@mail.msen.com) at March 28, 2020 11:58 PM

Collections.V3-dtl.883

Name: Collections-dtl.883
Author: dtl
Time: 28 March 2020, 4:33:55.690528 pm
UUID: e4123586-74d1-4edb-a358-46ce6d95631c
Ancestors: Collections-eem.882, Collections-ct.876

Remove WideString class>>initialize to followi up on an old to-do note from Andreas (see methods wtih @@@ comments for others).

Merge Collections-ct.876

by Dave Lewis (lewis@mail.msen.com) at March 28, 2020 11:57 PM

Compiler.V3-dtl.426

Name: Compiler-ct.426
Author: ct
Time: 28 March 2020, 1:22:49.51256 am
UUID: 761924e9-77a0-1144-9533-8a7c22c03332
Ancestors: Compiler-ct.424

Fixes a bug regarding decompilation of special selectors.

The following did not work before:

(OrderedCollection >> #asArray) decompile generate valueWithReceiver: {42} asOrderedCollection arguments: #(). "MessageNotUnderstood: OrderedCollection>>#Array=>Array"

See also this thread: http://forum.world.st/The-Inbox-EToys-ct-367-mcz-tp5105507p5114020.html

by Dave Lewis (lewis@mail.msen.com) at March 28, 2020 06:42 PM

Compiler.V3-dtl.424

For V3, merge change for Decompiler>>blockTo: (comment change only)

Name: Compiler-ct.424
Author: ct
Time: 26 March 2020, 5:19:54.71456 pm
UUID: 10f4405b-896c-3149-b76b-b9cd45ca5734
Ancestors: Compiler-nice.420

Fixes a bug regarding decompilation of FullBlockClosures

Sample to reproduce: [^ true] decompile should NOT return {[]}!

by Dave Lewis (lewis@mail.msen.com) at March 28, 2020 06:39 PM

Compiler.V3-dtl.421

Revert recent changes to Encoder>>litEndex:
Compiler recompileAll works again for V3.

by Dave Lewis (lewis@mail.msen.com) at March 28, 2020 06:33 PM

March 24, 2020

squeaksource.com

CryptographyPlugins-ul.22

- removed SHA256Plugin because SHA2Plugin covers SHA256

by Levente Uzonyi (leves@elte.hu) at March 24, 2020 12:11 AM

CryptographyPlugins-ul.21

Fixed several potential buffer overflow errors. And other minor tweaks.

DESPlugin:
- validate the size of the arguments of the primitives
- use primitive failure codes
- use #methodReturnBool: to return booleans

DSAPlugin:
- validate the numebr of arguments of the primitives
- always return after primitive failure (ignored argument validation)
- use primitive failure codes
- use #methodReturnBool: to return booleans

MD5Plugin:
- validate the size of the arguments of the primitives
- not a SmartSyntaxInterpreterPlugin anymore
- moved endianness check into #decode:sizeInIntegers:
- added inline pragmas
- use #byteSwap32 in #decode:sizeInIntegers:
- removed unused variable: swap
- changes may have broken simulation

SHA2Plugin:
- use #stSizeOf: instead of #slotSizeOf:
- assign a value to w in smalltalk in #sha256ProcessBuffer:updatingHash: and #sha512ProcessBuffer:updatingHash: to avoid the compiler complaining about unassigned variables and to help simluation

by Levente Uzonyi (leves@elte.hu) at March 24, 2020 12:11 AM

March 23, 2020

squeaksource.com

ParrotTalk-rww.44

Relocated CipherThunkMaker and EncoderThunk to ThunkStack.

by Robert Withers (robert.withers@protonmail.com) at March 23, 2020 08:57 PM

ParrotTalk-rww.44

Relocated CipherThunkMaker and EncoderThunk to ThunkStack.

by Robert Withers (robert.withers@protonmail.com) at March 23, 2020 08:57 PM