Planet Squeak Code

doWithIndex: => withIndexDo:

by Eliot Miranda (eliot.miranda@gmail.com) at January 20, 2021 04:47 AM

doWithIndex: => withIndexDo:

by Eliot Miranda (eliot.miranda@gmail.com) at January 20, 2021 04:45 AM

OSProcess 4.6.24
Cuis no longer has String>>asInteger so rewrite a few methods accordingly.

by Dave Lewis (lewis@mail.msen.com) at January 20, 2021 01:59 AM

OSProcess 4.6.24
Cuis no longer has String>>asInteger so rewrite a few methods accordingly.

by Dave Lewis (lewis@mail.msen.com) at January 20, 2021 01:58 AM

OSProcess 4.6.24
Cuis no longer has String>>asInteger so rewrite a few methods accordingly.

by Dave Lewis (lewis@mail.msen.com) at January 20, 2021 01:58 AM

Name: Collections-ul.922
Author: ul
Time: 19 January 2021, 1:28:51.784086 pm
UUID: 7d688371-2966-43d3-9000-3985ae9f6afa
Ancestors: Collections-nice.921

Fix off-by-one errors in OrderedCollection's #removeFirst: and #removeLast:. The argument of those methods can be zero.

by Dave Lewis (lewis@mail.msen.com) at January 19, 2021 11:52 PM

Revert the work-around in InstvarInterfaceExtractor>>nativeSend:numArgs: now that Collections-ul.922 fixes the issues with anOrderedColleciton removeLast: 0.

by Eliot Miranda (eliot.miranda@gmail.com) at January 19, 2021 08:33 PM

Add a test for the default Polynomial representation

by David Stes (stes@telenet.be) at January 19, 2021 01:30 PM

Fix off-by-one errors in OrderedCollection's #removeFirst: and #removeLast:. The argument of those methods can be zero.

by Levente Uzonyi (leves@caesar.elte.hu) at January 19, 2021 12:31 PM

- improved tests for OrderedCollection's #removeFirst: and #removeLast:.

by Levente Uzonyi (leves@caesar.elte.hu) at January 19, 2021 12:31 PM

ByteArray >> #bitXor: is part of the CryptographyCore package. Replace it with a simple loop to break the dependency.

by Levente Uzonyi (leves@elte.hu) at January 19, 2021 09:15 AM

Optimized HMAC by reducing the number of allocations:
1) allocate some extra bytes at the end of outerPadding to store the calculated hash there instead of concatenating two ByteArrays
2) use #hashStream:into: instead of #hashMessage: so that the same ByteArray can be used multiple times to store the calculated hash value
3) always use the same stream object, the argument of #hashStream:into:

If you have HMAC objects in your image, those will need to be recreated.

by Levente Uzonyi (leves@elte.hu) at January 19, 2021 09:10 AM

- some tests for PG3Scram

by Levente Uzonyi (leves@elte.hu) at January 19, 2021 06:33 AM

- added support for scram-sha-256 authentication
- show a more meaningful error message instead of 'Missing transition' when an error response is returned

by Levente Uzonyi (leves@elte.hu) at January 19, 2021 06:33 AM

Fix simulation of the ARMv5 code generator (increase method alignment to allow the entry alignment mask to be large enough, a la ARMv8).
Fix simulation of the V3 simulator (needs to implement getStackPointer for SmartSyntaxPlugin simulation).
Fix some speeling rorres in conemnts.

by Eliot Miranda (eliot.miranda@gmail.com) at January 19, 2021 01:00 AM

Fixes and supplements BecomeTest.

- Updated #testBecomeForwardIdentityHash which has been broken since Collections-eem.885. Note that the referenced patch to #becomeForward: was a breaking change which we should make sure to document in the final release notes.
- Added #testBecomeForwardIdentityAndHash to have a test for the classical primitiveArrayBecomeOneWay (primitive 72) again.
- Added #testBecomeForwardCopyIdentityHash to test both bindings of the copyHash argument indeed.

Please review! Fur further reference, see https://github.com/codefrau/SqueakJS/pull/117 where we were discussing the limitations/correctness of the current BecomeTest implementation.

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at January 18, 2021 11:55 AM

fix setScalarZero to set c zero instead of c scalarZero

by David Stes (stes@telenet.be) at January 17, 2021 06:56 PM

Add a BigInt Polynomial test

by David Stes (stes@telenet.be) at January 17, 2021 06:49 PM

Add event tracing mask, VM ticker stats, mixed arithmetic flag, and separate marking time.

by Eliot Miranda (eliot.miranda@gmail.com) at January 12, 2021 11:42 PM

Fixes a bug in resize grips (having #rigid targets) that are placed within layouted (usually #shrinkWrap) owners.

This makes all grips in resizable dialogs finally work as expected. I still consider this a #workaround to be improved.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 12, 2021 01:52 PM

Add a comment

by David Stes (stes@telenet.be) at January 12, 2021 01:45 PM

Removes the "breathing space" in a list-chooser's preferred extent. Thanks to Chris (cbc) for clarifying the issue with the cellInset in LazyListMorph.

Note that I do have a better solution for min/max/preferredExtent in combination with ProportionalLayout and LayoutFrame. Maybe now is the time to invest more time into that.

Also note that I will now investigate that 2-pixel-bug that causes the vertical scroll bar to remain after hitting the expand button in a dialog. It is related to the list's and input-field's borderWidth.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 12, 2021 01:09 PM

Fixes the issue where a dialog with flexible contents was not fully visible in small worlds. Thanks to Chris (cbc) for pointing this out!

(Note that I think that we need a better way for this combination of #fullBounds, #extent:, and #translatedToBeWithin: to make sure that a morph is visible in the world. May be also useful for other windows. Not sure about this variation with #center: though.)

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 12, 2021 01:04 PM

Better estimated size:
- includes default cellInsert for listMorph
- includes boundary line size
Also tries to size for width of strings - at least, the first 15 samples.

January 11, 2021 07:45 PM

Set the initial extent of the ListChooser to not require resizing the dialog for reasonable sized lists (i.e., expand it so you can see all of the list).
At the small end, still object MT's suggested size; at the large end, will not take up more the 3/4ths of the world size.

by Chris Cunningham (cunningham.cb@gmail.com) at January 11, 2021 07:41 PM

Let the width of List Chooser try to accomodate better to the actual size of the string.

by Chris Cunningham (cunningham.cb@gmail.com) at January 11, 2021 07:40 PM

Minor fix in dialog to not hide the world's docking bar.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 11, 2021 04:39 PM

To pluggable dialogs, add an expand button if the model supports #preferredExtent. Also tweaks a list-chooser's initial extent show 5 to 15 items which each about 10 to 20 characters, depending on the list's content.

This commit addresses the concerns raised in ToolBuilder-Morphic-cbc.270 (inbox).
See http://forum.world.st/The-Inbox-ToolBuilder-Morphic-cbc-270-mcz-tp5126045.html

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 11, 2021 04:36 PM

Provide ImageFileHeader class>>headerAndFlagsFrom: to read header from an image file and remember the endianness flag used during the read. Intended to enable ImageSnapshot to be instantiated from an image file as well as directly from a VM primitive.

by David T. Lewis (lewis@mail.msen.com) at January 10, 2021 08:38 PM

Update a method comment

by Dave Lewis (lewis@mail.msen.com) at January 10, 2021 07:46 PM

Provide ImageSnapshot class>>fromStream: to allow instantiation from a saved image file in addition to the normal direct instantiation via VM primitive.

by Dave Lewis (lewis@mail.msen.com) at January 10, 2021 07:35 PM

Import classes from sourceforge

by David Stes (stes@telenet.be) at January 10, 2021 07:03 PM

Harumph; sqMemoryAccess.h& sqPlatformSpecific.h must be included in that order :-(

by Eliot Miranda (eliot.miranda@gmail.com) at January 08, 2021 07:32 AM

Have the Cogits include sqPlatformSpecific.h for solaris.

by Eliot Miranda (eliot.miranda@gmail.com) at January 08, 2021 07:22 AM

Time to bring the ARMv8 work back into Cog/VMMaker under full MIT.

by Eliot Miranda (eliot.miranda@gmail.com) at January 08, 2021 07:02 AM

Time to bring the ARMv8 work back into Cog/VMMaker under full MIT.

by Eliot Miranda (eliot.miranda@gmail.com) at January 08, 2021 06:59 AM

Bow to LLP64 and type ioMicroMSecs as unigned int.

by Eliot Miranda (eliot.miranda@gmail.com) at January 08, 2021 12:58 AM

Name: System-dtl.1210
Author: dtl
Time: 6 January 2021, 4:59:31.092128 pm
UUID: a704d749-4986-46f9-aa2f-faa8bb4ea5e1
Ancestors: System-eem.1207

Provide vmParameterAt:ifAbsent: to handle primitive failure on reading VM parameters.
Supply default parameter values to mock possibly missing elements in the parameters array.

by Dave Lewis (lewis@mail.msen.com) at January 08, 2021 12:32 AM

Newspeak Cogit:
Reimplement warnMultiple:selectors: given printf support for #stderr as a parameter and %.*s variable string widths.

by Eliot Miranda (eliot.miranda@gmail.com) at January 07, 2021 11:45 PM

Extend to support %.*s string widths, where the preceding parameter gives the width of the string.

by Eliot Miranda (eliot.miranda@gmail.com) at January 07, 2021 10:43 PM

x86/x86_64 Cog MTVM:
Add the code to handle the lock already having the right value to x86/x86_64 generateLowLevelTryLock:.
Extend handleCompareAndSwapSimulationTrap: to simulate x86/x86_64 cmpxchg (add failedComparisonRegisterAccessor to the mix).
Always use a REX prefix for SETE.
Categorize all processor specific opcode generators under concretize processor-specific

Still the x86_64 low-level lock dfoesn't simulate correctly. I suspect that setting the flags word doesn't actually set the flags in the plugin.

by Eliot Miranda (eliot.miranda@gmail.com) at January 07, 2021 10:05 PM

MTVM:
Extend CompareAndSwapSimulationTrap with the failedComparisonRegisterAccessor to to CMPXCHG correctly on x86/x86_64.

Print the direction flag on x86/x86_64.

Apply the clone: => cloneObject: refactoring.

by Eliot Miranda (eliot.miranda@gmail.com) at January 07, 2021 10:01 PM

Include some more include files in the Cogits that are included by sq.h.
Rename clone: to cloneObject: to avoid the clash with Unix pthread.h/sched.h's definition of clone, a variant of fork.

by Eliot Miranda (eliot.miranda@gmail.com) at January 07, 2021 06:43 PM

Provide vmParameterAt:ifAbsent: to handle primitive failure on reading VM parameters.
Supply default parameter values to mock possibly missing elements in the parameters array.

by David T. Lewis (lewis@mail.msen.com) at January 06, 2021 09:59 PM

Backports menu-builder fix from 6.0alpha.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 10:14 AM

Improve robustness against erroneous menu builders.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 10:12 AM

Backports some fixes from 6.0alpha:
- choose from value list works again
- annotation pane works again

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 10:00 AM

Backports some fixes from 6.0alpha:
- find/replace in text fields works again
- auto-indent on CR works again

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 09:58 AM

Merges ToolBuilder-MVC-ct.60, which got shadowed by ToolBuilder-MVC-TheresaHMartenK.60 last June.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 09:54 AM

Fixes auto-indent bug in MVC text editors by supporting type-aheads with size > 1 outside the context of copy-paste (and similar) in #zapSelectionWithCompositionWith:.

Note that dynamic character composition is not supported in Morphic's TextEditor, I suppose. We might want to add that, too. Yoshiki added this feature in 2007 only to ST80/MVC.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at January 06, 2021 09:33 AM

Cogit:
More cleanup given that sqVirtualMachine.h reveals the direct interpreter API to SQUEAK_BUILTIN_PLUGIN which is alas defined for cogit.c. Hence the conflicts between sqVirtualMachine.h and cointerp.h force us to not include sq.h in cogit.c. Maybe this is worse than the disease of sqVirtualMachine.h's API being inaccurate. But ione step at a time.

MTVM:
Smilaiton time recording of instructions in tryLockVMOwnerTo: while debugging x86_64's implementation.

by Eliot Miranda (eliot.miranda@gmail.com) at January 06, 2021 12:07 AM

OIbserve stricter type compatibility now that sqVirtualMachine.h's definitions are visible to internal plugins.

by Eliot Miranda (eliot.miranda@gmail.com) at January 05, 2021 11:27 PM

Oops! Remember to nuke the unused setFileAccessCallback:

by Eliot Miranda (eliot.miranda@gmail.com) at January 05, 2021 08:21 PM

Small fix for tests

concretizeAt: does not answer the instruction size but the next address

by nicolas cellier (nicolas.cellier.aka.nice@gmail.com) at January 02, 2021 12:42 PM

correction of seaside component config file

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 12:06 PM

Seaside components config

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 12:00 PM

Seaside components commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 12:00 PM

test config

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 11:38 AM

jUST TEST COMMIT

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 11:18 AM

bootstrap commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 10:27 AM

seaside commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 10:26 AM

Zinc commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 10:25 AM

1st commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 10:23 AM

1st commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 10:22 AM

Seaside config commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 08:23 AM

Zinc COnfig commit

by Yousef (eng.yousef416@yahoo.com) at January 02, 2021 08:21 AM

Spur Cogit:
Fix the broken value range checking code wot I wrote for signed bits array at:put:.

Simulation:
Fix mapping of break pcs on code compaction when break pc is an array of pcs.

by Eliot Miranda (eliot.miranda@gmail.com) at January 02, 2021 05:13 AM

Add a FloatTest to compare all literal float contants in methods with those from a fresh compile, failing if any differ. Add a utility method to answer teh set of methods that cause the test to fail (FloatTest>>methodsMaybeContainingBrokenCompiledConstants).

by Eliot Miranda (eliot.miranda@gmail.com) at January 01, 2021 08:17 PM

Oops. remember to include the meta part of the test.

by Eliot Miranda (eliot.miranda@gmail.com) at January 01, 2021 07:29 PM

Add a test to check Float's constants. Add a utility (e.g for package prologs) to answer whether the test passes. At least in my image Float fmax had become NaN.

by Eliot Miranda (eliot.miranda@gmail.com) at January 01, 2021 07:24 PM

Tests simulation of #perform:... primitives 83, 84, and 100. Complements Kernel-ct.1367.

Depends indeed not only on KernelTests-tonyg.381 but also on KernelTests-ct.375, it would be nice if we could get the latter merged soon, this has already been causing too many merge conflicts in the past. :-)

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at January 01, 2021 06:59 PM

Fixes simulation of the #perform:... primitives 83, 84, and 100 for all edge cases. If the primitive is called with the wrong arguments, the primitive must fail but not the simulator.

For further reference, see also the implementation of primitive 188 (primitiveExecuteMethodArgsArray) just a few lines below. :-)

Tests are following right now, I am looking forward to your review!

January 01, 2021 06:53 PM

another commit 1st try

by Yousef (eng.yousef416@yahoo.com) at January 01, 2021 08:01 AM

1st commit

by Yousef (eng.yousef416@yahoo.com) at January 01, 2021 07:44 AM

fifth

by Yousef (eng.yousef416@yahoo.com) at December 31, 2020 11:12 PM

can be forth config commit

by Yousef (eng.yousef416@yahoo.com) at December 31, 2020 11:11 PM

can be 3rd commit

by Yousef (eng.yousef416@yahoo.com) at December 31, 2020 11:11 PM

Provide vmParameterAt:default: to handle primitive failure on reading VM parameters.
Supply default parameter values to mock possibly missing elements in the parameters array.
Remove two unnecessary isRunningCog checks and an ifNotNil:
Remove inappropriate halt in sendMouseWheelEvents:
Let supportsMultipleBytecodeSets and supportsReadOnlyObjects work on any VM.

by David T. Lewis (lewis@mail.msen.com) at December 31, 2020 09:27 PM

Name: Collections-nice.921
Author: nice
Time: 30 December 2020, 12:46:21.328377 am
UUID: f66c5960-3b41-4815-ab2c-ce42061fb43e
Ancestors: Collections-eem.920

Fix buggish float array 32<->64 conversions...

Hell, how could I publish that?
Il ne faut pas confondre vitesse et précipitation...

by Dave Lewis (lewis@mail.msen.com) at December 31, 2020 03:48 PM

VMMaker 4.19.5
Refactoring - move duplicated variables and methods from StackInterpreter and ContextInterpreter up to Interpreter.
Remove stack interpreter support methods from ContextInterpreter, they are only sent from NewObjectMemory and StackInterpreter.
Move duplicated stack bytecodes methods up to Interpreter.
Move primitiveSetGCSemaphore up to Interpreter.
Move duplicated common selector sends methods up to Interpreter.
Move duplicated bitbit support methods up to Interpreter.
Move duplicated array primitive support methods up to Interpreter.
Move duplicated arithmetic primitive support methods up to Interpreter and remove unsent checkIntegerResult:..
Move duplicated I/O primitive methods up to Interpreter.

by David T. Lewis (lewis@mail.msen.com) at December 31, 2020 03:31 PM

VMMaker 4.19.4
The global session identifier must be updated when the intepreter resumes into a new image via primitiveResumeFromSnapshot.
Refactoring - move duplicated variables and methods from StackInterpreter and ContextInterpreter up to Interpreter.

by David T. Lewis (lewis@mail.msen.com) at December 31, 2020 03:30 PM

fix error in steps calculation

by Thiago da Silva Lino (thiagolino@yahoo.com.br) at December 31, 2020 10:36 AM

Enable using a WordArray or other RawBitsArray species as simulation memory.
This implies using byteSize at a few places.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 31, 2020 09:08 AM

A few fixes for the VM tests
- enable using a WordArray as simulation memory
- concretizeAt: does not answer the instruction size but the next address

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 31, 2020 09:06 AM

ST80 doesn't seem to depend on Monticello anymore.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 28, 2020 01:55 PM

Proposal: Dispatch request notifications via ToolSet before opening an interactive dialog window. This should give us better feedback in non-interactive environments such as CI jobs or production contexts (at the moment, CI only hangs silently).

Questions: Is it okay to depend on the System, Tools, and CommandLineTools from Monticello? Are there further exceptions that should be dispatched equally?

December 28, 2020 11:48 AM

Add displayingProcess for Pharo portability

December 28, 2020 06:57 AM

Improve portability to Pharo, tidy up

December 28, 2020 04:48 AM

page 56 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 28, 2020 02:36 AM

page 56 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 28, 2020 02:36 AM

page 56 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 28, 2020 02:36 AM

More tags supported. Validated with over 700 audiobooks (no music/movies, so that metadata section is not validated).

December 28, 2020 02:28 AM

Successfully parsed the nearly 1,000 or so MP3 files I have, successfully enough for my purposes.
Fairly extensive overhaul; works with ID3v2.2, ID3v2.3, ID3v2.4, most frame types, and several very weird variations that do not follow the standard (yuck).

December 27, 2020 10:49 PM

Proposal: Isolate alpha channel when printing a named color. This allows it to reuse the color name even for translucent color.

Example:
(Color red alpha: 0.4) printString
Output (new):
'Color red alpha: 0.4'
Output (old):
'(TranslucentColor r: 1 g: 0.0 b: 0.0 alpha: 0.4)'

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 10:16 PM

Makes color printing more consistent. #printString does not add brackets, #storeString does.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 10:15 PM

Isolate alpha channel when printing a named color. This allows it to reuse the color name even for translucent color.

Example:
(Color red alpha: 0.4) printString
Output (new):
'(Color red alpha: 0.4)'
Output (old):
'(TranslucentColor r: 1 g: 0.5 b: 0.0 alpha: 0.4)'

Note: the implementation preserves parentheses and Color transparent printString.

Thanks to Christoph Thiede (ct) for the original proposal.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 10:14 PM

Merge Graphics-mt.441, Graphics-ct.439, Graphics-ct.430, Graphics-ct.425, Graphics-ct.418

Graphics-mt.441:
Some clean up around DisplayTransform. Thanks to Stephan Lutz (stlu)!

Graphics-ct.439:
Fixes transparency handling in Color >> #negated. Negating a color should not remove its alpha channel.

Graphics-ct.430:
Miscellaneous minor refactoring

- Use #ifError: and correct deprecated [:argWithoutBar] block syntax (see Compiler-ct.417)
- Fix a documetation typo
- Simplify some conditions

Graphics-ct.425:
[very minor] Use already defined magic number instead of hard-coded duplication

Graphics-ct.418:
Add Color>>#veryMuchDarker

We also have #veryMuchLighter, so balance this out :)

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 09:35 PM

[very minor] Use already defined magic number instead of hard-coded duplication

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 09:13 PM

Miscellaneous minor refactoring

- Use #ifError: and correct deprecated [:argWithoutBar] block syntax (see Compiler-ct.417)
- Fix a documetation typo
- Simplify some conditions

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 09:12 PM

Fixes transparency handling in Color >> #negated. Negating a color should not remove its alpha channel.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 09:00 PM

Adds regression test for Graphics-ct.439 (Color >> #negated).

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 08:58 PM

End of the #World as we know it. Package postscript only. About two years ago we eliminated all dependencies on global World. However, the global binding has remained functional on the theory that some external packages might still expect it.

It is time to get rid of the binding now because:
- It gives the false impressing that the World is still in use as a global
- If a package is loaded that does need World, then "Smalltalk at: #World put: Project current world" will restore prior behavior.

by David T. Lewis (lewis@mail.msen.com) at December 27, 2020 07:32 PM

Add tests for new alpha compositing rules.

See https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/505

Those tests require a newer VM (or at least a newer BitBlt plugin - post VMMaker.oscog-nice.2909).

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 06:29 PM

Add tests for new alpha compositing rules.

See https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/505

Those tests require a newer VM (or at least a newer BitBlt plugin - post VMMaker.oscog-nice.2909).

This is the 2nd attempt with typo corrected:
Uncaled => Unscaled
Hence, GraphicsTests-nice.56 can be thrown away.

December 27, 2020 06:13 PM

Provide comment and named accessors for new alpha BitBlt compositing combination rules.

Those new rules are described here:
https://github.com/OpenSmalltalk/opensmalltalk-vm/issues/505

They require a new VM, or at least a new BitBlt plugin posterior to VMMaker.oscog-nice.2909

December 27, 2020 06:10 PM

Complexify the rule for generating hex literal constants when more intellegible than decimal.

This is useful for having a chance to decipher generated code for bit tricks.

This replaces VMMaker.oscog-nice.2911 that screwed things up (hex generates the radix 16r).
VMMaker.oscog-nice.2911 should be thrown away.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 03:21 PM

Name: Compiler-eem.453
Author: eem
Time: 26 December 2020, 7:30:51.52735 pm
UUID: 1a90195c-0773-4f16-bc29-08b718ad4d66
Ancestors: Compiler-eem.452

Fix a spelling error, preceed => precede, etc

by Dave Lewis (lewis@mail.msen.com) at December 27, 2020 02:30 PM

Simplify the rule for generating hex literal constants when more intellegible than decimal.

This is useful for having a chance to decipher generated code for bit tricks.

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 27, 2020 02:28 PM

Fix a spelling error, preceed => precede, etc

by Eliot Miranda (eliot.miranda@gmail.com) at December 27, 2020 03:30 AM

Raise cog v3 VM compatibility to 1.17

Use a few forward compatibility messages to do so...

This is not ideal for double byte and double word arrays which are not supported.
Neither for word and float32 arrays that lack appropriate specialObject index...

The goal is to have the squeak.cog.v3 to at least compile the new file and socket plugin versions...

by Nicolas Cellier (nicolas.cellier.aka.nice@gmail.com) at December 26, 2020 11:00 PM

Name: System-eem.1207
Author: eem
Time: 22 December 2020, 6:25:35.535409 pm
UUID: 307e584e-0a0c-422b-8f92-2acd43a29ff7
Ancestors: System-dtl.1206

Update vmParameterAt: doc with the event trace mask.
Fix ResourceManager>>#convertMapNameForBackwardcompatibilityFrom: for the 10.x => 11.x MacOS version name issue.

by Dave Lewis (lewis@mail.msen.com) at December 25, 2020 11:21 PM

Bug fixes (with syncSafeIntegers - among others, moved this logic to ByteArray)
Factored out Constants.
Added reading of Xing header.
Extended header fields read.

December 22, 2020 05:36 PM

empty log message

December 22, 2020 05:04 PM

Moving to Pi-4 from Pi-3 to continue development with two HDHomeRun devices with a total of 6 tuners, 2 of which are 4K.

The code is at an Alpha stage. I have many of the elements in testing including a working priority filter.

by John-Reed Maffeo (jrmaffeo@gmail.com) at December 22, 2020 03:30 PM

DoItFirst>>parse: fix a shadowed variable and provide better method comment

by David T. Lewis (lewis@mail.msen.com) at December 22, 2020 04:34 AM

Minor changes to allow package to load in old Squeak images

by David T. Lewis (lewis@mail.msen.com) at December 21, 2020 02:38 PM

Update platformName for SqueakJS 1.0

by Vanessa Freudenberg (vanessa@codefrau.net) at December 21, 2020 06:24 AM

VMMaker 4.19.3
Let primitiveResumeFromSnapshot handle a changed image format for the resumed snapshot object.
Install and update an appropriate primitive table dynamically on entry to the interpreter loop.
Reset the image format number in the interpreter when resuming a snapshot object.
These changes enable e.g. a Squeak 4.6 image to resume into a Squeak 3.6 snapshot and vice versa.

by David T. Lewis (lewis@mail.msen.com) at December 21, 2020 02:24 AM

Allow any lengthOfLine, fix bugs

December 20, 2020 01:36 PM

Provide workaround for Squeak4.6 not responding to localOffsetSeconds

by David Stes (stes@telenet.be) at December 20, 2020 01:03 PM

Few improvements for release of Aida 6.8

by David Stes (stes@telenet.be) at December 20, 2020 11:53 AM

page 56 WIP

by Fujio Tanabe (ijktanabe@gmail.com) at December 20, 2020 06:22 AM

page 56 WIP

by Fujio Tanabe (ijktanabe@gmail.com) at December 20, 2020 06:21 AM

page 56 WIP

by Fujio Tanabe (ijktanabe@gmail.com) at December 20, 2020 06:21 AM

Ensure integral nanoseconds when printing DateAndTime to prevent error in e.g.
(DateAndTime fromSeconds: 3124074224.123456789s) printString

by David T. Lewis (lewis@mail.msen.com) at December 20, 2020 12:57 AM

Refactor BotPlayer, comments cleanup

December 15, 2020 05:58 PM

Fix several osVersion tests on mac OS that fail for Big Sur (11.x). The old code read
'10*' match: osVersion
which fails on Big Sur, e.g. '1100.1', '1101.0'.
The new code tests for the index of the period being > 4. Arguably the old code is obsolete.

I'm guessing as to the old format. If in fact it was only three digits then my code is wrong, and should be e.g. Smalltalk osVersion includes: $.
Please let me know.

One method SimplifiedChineseEnvironment class>>#inputInterpreterClass is simplified, eliminating redundant tests.

by Eliot Miranda (eliot.miranda@gmail.com) at December 15, 2020 02:53 AM

page 50 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 14, 2020 09:19 AM

page 50 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 14, 2020 09:19 AM

Mpeg3Plugin sample read routines expect word arrays,not pointer arrays. Became an issue recently due to Slang changes?

by tim Rowledge (tim@rowledge.org) at December 14, 2020 05:10 AM

Name: Kernel-dtl.1363
Author: dtl
Time: 8 December 2020, 4:13:03.533263 pm
UUID: dc65867e-37e6-4295-aa44-ef26c62ba250
Ancestors: Kernel-eem.1362

Let Delay class>>startup invoke DoItFirst class>>reevaluateDebug. If a DoItFirst command line option has requested a debugger, then invoke it now after Delay startUp processing..

by Dave Lewis (lewis@mail.msen.com) at December 13, 2020 11:10 PM

Name: System-mt.1196
Author: mt
Time: 9 December 2020, 11:08:33.726163 am
UUID: 93f95297-e0dd-43a5-84a8-16fa9ea4c6e2
Ancestors: System-dtl.1195

Removes duplication for author initials "md".

by Dave Lewis (lewis@mail.msen.com) at December 13, 2020 11:08 PM

Name: System-dtl.1195
Author: dtl
Time: 8 December 2020, 4:10:40.495367 pm
UUID: 88a0ce82-643e-47c4-83e9-27404ce5f71f
Ancestors: System-eem.1194

Add DoItFirst to be the first thing in the system startup list, processing certain command line options prior to any additional image initialization.
DoItFirst image arguments:
--doit argumentlist "evaluate each argument as a doIt expression"
--evaluate arg "evaluate arg, print result then exit"
--file filename "evaluate contents of filename, print result then exit"
--filein filelist "file in each file named in fileList"
--cwd path "set FileDirectory defaultDirectory to path prior to evaluating other options"
--debug "enter a debugger as soon as possible in the startUp processing"
--help "print this message"
Some arguments have single character synonyms, -f is a synonym for --file, -d for --doit
A single '-' may be used instead of '--', -help is interpreted as --help

by Dave Lewis (lewis@mail.msen.com) at December 13, 2020 09:55 PM

page 48 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 07:59 AM

page 48 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 07:59 AM

page 48 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 05:40 AM

page 48 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 05:40 AM

Enable removal of preference for 'Read document at startup'.

Simplify code such that #readDocumentAtStartup is assumed always true and document specifiers begining with '-' are not treated as start documents. The actual preference setting is no longer referenced.

Assumptions: if the start document resource name starts with '-' then it is unlikely to be a document. For the case of the first argument appearing to be a document but the user wishing to use it otherwise, the '--' token (or any other argument beginning with '-') may be used in the command line to protect the argument from evaluation. This allows the traditional start document processing to work normally in the general case, and allows start document processing to be bypassed from the command line if desired. No preference setting is required.

The #readDocumentAtStartup may be removed in a future update.

December 13, 2020 04:53 AM

page 40 WIP

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 12:24 AM

page 40 WIP

by Fujio Tanabe (ijktanabe@gmail.com) at December 13, 2020 12:23 AM

Fixes code styling in change sorters.

Funnily, #aboutToStyle: was already implemented on ChangeSorter, only the toolbuilder method needed a small update ... :-)

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at December 12, 2020 10:33 PM

Initial commit

December 09, 2020 03:42 PM

Adds a simple (but working) inspector for CharacterSet.

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at December 09, 2020 03:22 PM

Removes duplication for author initials "md".

by Marcel Taeumel (marcel.taeumel@hpi.uni-potsdam.de) at December 09, 2020 10:08 AM

Let DoItFirstTest>>startup invoke DoItFirst>>reevaluateCwd. If a DoItFirst command line option has specified a new FileDirectory default, apply it now.

by David T. Lewis (lewis@mail.msen.com) at December 08, 2020 09:13 PM

Let Delay class>>startup invoke DoItFirst class>>reevaluateDebug. If a DoItFirst command line option has requested a debugger, then invoke it now after Delay startUp processing..

by David T. Lewis (lewis@mail.msen.com) at December 08, 2020 09:13 PM

Add DoItFirstTest to verify command line parsing for DoitFirst and confirm mapping of command line tokens to known options.

by David T. Lewis (lewis@mail.msen.com) at December 08, 2020 09:12 PM

Add DoItFirst to be the first thing in the system startup list, processing certain command line options prior to any additional image initialization.
DoItFirst image arguments:
--doit argumentlist "evaluate each argument as a doIt expression"
--evaluate arg "evaluate arg, print result then exit"
--file filename "evaluate contents of filename, print result then exit"
--filein filelist "file in each file named in fileList"
--cwd path "set FileDirectory defaultDirectory to path prior to evaluating other options"
--debug "enter a debugger as soon as possible in the startUp processing"
--help "print this message"
Some arguments have single character synonyms, -f is a synonym for --file, -d for --doit
A single '-' may be used instead of '--', -help is interpreted as --help

by David T. Lewis (lewis@mail.msen.com) at December 08, 2020 09:10 PM

page 35C done

by Fujio Tanabe (ijktanabe@gmail.com) at December 08, 2020 12:22 PM

page 35C done

by Fujio Tanabe (ijktanabe@gmail.com) at December 08, 2020 12:22 PM

page 35 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 08, 2020 11:04 AM

page 35 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 08, 2020 11:04 AM

Tests Collections-ct.922 (Dictionary >> #at:put:during:).

December 07, 2020 05:47 PM

Proposal: Implements #add:put:during: as execution around method on Dictionary. This logic is not absolutely trivial, so I would not like to reimplement it again every time I need to build an execution around setter.

December 07, 2020 05:46 PM

page 34 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 07, 2020 09:47 AM

page 34 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 07, 2020 09:47 AM

Add unit tests to document parsing of command line tokens to option argument actions.

by Dave Lewis (lewis@mail.msen.com) at December 07, 2020 12:04 AM

Update class comment

by Dave Lewis (lewis@mail.msen.com) at December 06, 2020 06:40 PM

Re-enables handling of #launchDrop events when the VM is configured as a singleton and has been invoked again, though currently only handled with a rudimentary fallback implementation.

See http://forum.world.st/Changeset-Enhanced-integration-of-drag-n-drop-from-host-tp5123857p5124332.html. Thanks to Jakob (jr) for the feedback!

Reuploaded as replacement for Morphic-ct.1715. Thanks also to Marcel for the feedback! :-)

by Christoph Thiede (Christoph.Thiede@student.hpi.uni-potsdam.de) at December 06, 2020 05:03 PM

Implemented JsonObject >> #respondsTo: which returns true for all setters and getters of already defined fields. The code works the same way #doesNotUnderstand: does.

by Levente Uzonyi (leves@elte.hu) at December 06, 2020 04:47 PM

- make it possible to access the field named 'value' of a PG3Row by sending #value to it. Just like how #name works

by Levente Uzonyi (leves@elte.hu) at December 06, 2020 04:40 PM

Fix --filein which needs FileDirectory initialization. Merge DoItFirst-System-Support-dtl.14 which was accidentally skipped in the last commit.

by Dave Lewis (lewis@mail.msen.com) at December 06, 2020 04:40 PM

page 33 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 04:24 PM

page 33 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 04:24 PM

Reorganize parsing to support single character shortcuts, e.g. -e for --evaluate.
Add a command line option --file (or -f) for evaluate the contents of a file, print string result.

by Dave Lewis (lewis@mail.msen.com) at December 06, 2020 01:58 PM

page 30 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 08:16 AM

page 30 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 08:15 AM

page 28 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 07:33 AM

page 28 done

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 07:33 AM

done 28

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 05:31 AM

done 28

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 05:31 AM

done 026

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 05:01 AM

done 026

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 05:00 AM

empty log message

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 01:13 AM

empty log message

by Fujio Tanabe (ijktanabe@gmail.com) at December 06, 2020 01:11 AM

Assume that we have both bash and tar available. Compress the backups into a tgz archive file and deleted the inage and changes files. Run this in a background unix process (not OSProcess, just a shell command line with &).

by Dave Lewis (lewis@mail.msen.com) at December 04, 2020 02:26 AM

Name: System-eem.1194
Author: eem
Time: 2 December 2020, 9:58:22.669121 pm
UUID: e3212f4a-d84c-4e35-b05e-ab5e16af91c0
Ancestors: System-tonyg.1193

When computing SystemNavigation>>allSentMessagesWithout: use anyAndAllSelectorsDo: which does a much better job than the existing code, and uses (indirectly) Scanner>>isMessageSelector: which provides a single point of definition to maintain cnsistency across tools, etc.

by Dave Lewis (lewis@mail.msen.com) at December 03, 2020 11:19 PM

Name: Kernel-eem.1362
Author: eem
Time: 2 December 2020, 9:54:15.943794 pm
UUID: c9a9d41d-b1f3-443d-94bd-916e18b2ccba
Ancestors: Kernel-eem.1361

Provide CompiledMethod>>anyAndAllSelectors[Do:] that includes unsent symbols that might be selectors. e.g. HaloMorph>>#addRotateHandle: really does send
#startRot:with: and #doRot:with:, albeit indirectly via perform:.

by Dave Lewis (lewis@mail.msen.com) at December 03, 2020 11:17 PM

Name: Compiler-eem.452
Author: eem
Time: 2 December 2020, 9:41:04.063795 pm
UUID: 4799ba54-ed87-4188-a653-e36499d92994
Ancestors: Compiler-eem.451

Provide an acceptably accurate means to test if a SYmbol is probably a message selector, Scanner class>>isMessageSelector:.

by Dave Lewis (lewis@mail.msen.com) at December 03, 2020 11:15 PM

Name: Collections-eem.920
Author: eem
Time: 2 December 2020, 9:49:01.130251 pm
UUID: 3767683c-33f6-4874-9e6c-83719db04fe3
Ancestors: Collections-mt.919

Add Symbol>>isMessageSelector, implemented in terms of Scanner class>>isMessageSelector:

by Dave Lewis (lewis@mail.msen.com) at December 03, 2020 11:14 PM

When computing SystemNavigation>>allSentMessagesWithout: use anyAndAllSelectorsDo: which does a much better job than the existing code, and uses (indirectly) Scanner>>isMessageSelector: which provides a single point of definition to maintain cnsistency across tools, etc.

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:58 AM

In the browsers use Compiledmethod>>anyAndAllMessages instead of CompiledMethod>>messages. false positives are more tolerable tnan omitted selectors sent via perform: et al.

e.g. HaloMorph>>#addRotateHandle: really does send
#startRot:with: and #doRot:with:, albeit indirectly via perform:.

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:56 AM

Provide CompiledMethod>>anyAndAllSelectors[Do:] that includes unsent symbols that might be selectors. e.g. HaloMorph>>#addRotateHandle: really does send
#startRot:with: and #doRot:with:, albeit indirectly via perform:.

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:54 AM

Add Symbol>>isMessageSelector, implemented in terms of Scanner class>>isMessageSelector:

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:49 AM

Provide an acceptably accurate means to test if a SYmbol is probably a message selector, Scanner class>>isMessageSelector:.

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:41 AM

Allow the TestRunner's categories to be sorted, even if this doesn't persist past running the selected test(s). But for finding a category it can be more convenient than filtering.

by Eliot Miranda (eliot.miranda@gmail.com) at December 03, 2020 05:39 AM

Deprecate #totalSeconds and #millisecondClockValue in DateAndTime and refer senders to the canonical implementations in Time.

by David T. Lewis (lewis@mail.msen.com) at December 03, 2020 02:44 AM

Deprecate #totalSeconds and #millisecondClockValue in DateAndTime and refer senders to the canonical implementations in Time.

by David T. Lewis (lewis@mail.msen.com) at December 03, 2020 02:43 AM

Eliminate the only known use of DateAndTime class>>totalSeconds, use the proper method in class Time instead.

by David T. Lewis (lewis@mail.msen.com) at December 03, 2020 02:21 AM

Nicer code for a couple of Duration methods...
I'm doing this to help me remember 5 seconds wait et al...

by Eliot Miranda (eliot.miranda@gmail.com) at December 02, 2020 07:42 PM

Adds a development 2.1 version for Refactoring Tools

December 02, 2020 11:45 AM

Cogit: extend the assert in cogMethodContaining: to handle interruption at backward branches. Rename ceCheckForInterrupts to ceCheckForInterrupt to match Cogit's ceCheckForInterruptTrampoline inst var.

SoundPlugin/SmartSyntaxPlugin Slang:Fix a compile-time warning due to a bad cast of firstIndexableField: for WordsOrBytes parameters.
Nuke usobsoleted generators (ccgLoad:expr:asCharPtrFrom: superceded by ccgLoad:expr:asCharPtrFrom:andThen:, etc).
Avoid the unnecessary cast of the void * return type of firstIndexableField to parameters.

Simulaiton:
Add SignedCArray and coercion between it and CArray.

by Eliot Miranda (eliot.miranda@gmail.com) at December 02, 2020 03:54 AM

Fix a bug in SoundRecorder so that the following works:

[:r| r resumeRecording. (Delay forSeconds: 5) wait. r playback] value: SoundRecorder new

Arguably this should work too:

[:r| r startRecording. (Delay forSeconds: 5) wait. r playback] value: SoundRecorder new

i.e. startRecording should set paused to false. But this is a change in API that could affect clients so I'll oput the suggestion out there and leave it to others to judge.

by Eliot Miranda (eliot.miranda@gmail.com) at December 02, 2020 12:41 AM

Comment and category name tweaks

by Dave Lewis (lewis@mail.msen.com) at December 01, 2020 01:38 AM

MTVM: fix preemptDisowningThread; marryFrame:SP: expects top-of-stack to be the Smalltalk top-of-stack, not a pushed instructionPointer.

by Eliot Miranda (eliot.miranda@gmail.com) at November 30, 2020 08:09 PM

...and reenterInterpreter is no longer an inst var of CogVMThread

by Eliot Miranda (eliot.miranda@gmail.com) at November 30, 2020 05:59 PM

Oops; fix CogVMThread typedef.

by Eliot Miranda (eliot.miranda@gmail.com) at November 30, 2020 05:57 PM

Add a guard to FileDirectory class>>requestDropDirectory:. Every time I connect my headphone buds I get an MNU because primDropRequestFileName: answers nil. This squashes the MNU.

by Eliot Miranda (eliot.miranda@gmail.com) at November 29, 2020 06:24 PM

MTVM: Eliminate some compiler warnings. Fix a slip in the assert in cogMethodContainng:
In simulation get register state correct on proceeding from ioWaitOnOSSemaphore:

by Eliot Miranda (eliot.miranda@gmail.com) at November 29, 2020 06:54 AM

Simulaiton:
Symbolic names for addresses on the rump C stack.
More asserts tracking c/processor stack pointers in the MT VM.

by Eliot Miranda (eliot.miranda@gmail.com) at November 25, 2020 07:25 AM