Squeak-dev September 2011

squeak-dev@lists.squeakfoundation.org

64 participants
208 discussions

Re: [squeak-dev] The Inbox: Tools-ul.378.mcz
by Eliot Miranda 09 Sep '11

09 Sep '11

Levente, I object :) I find the findBinary:do:ifNone: extremely ugly and much prefer the findNearbyBinaryIndex: usage. Would you mind if we reverted to that? On Thu, Sep 8, 2011 at 2:04 PM, <commits(a)source.squeak.org> wrote: > A new version of Tools was added to project The Inbox: > http://source.squeak.org/inbox/Tools-ul.378.mcz > > ==================== Summary ==================== > > Name: Tools-ul.378 > Author: ul > Time: 8 September 2011, 10:57:10.107 pm > UUID: 0262d60e-4ceb-b246-8d3f-96424d2aa1ce > Ancestors: Tools-eem.377 > > Simplified DebuggerMethodMap >> #rangeForPC:contextIsActiveContext:. > - copy and sort the associations in a bit simpler and probably faster way > - removed unnecessary checks (#isNil after non-nil assignment) > - uses Array instead of SortedCollection :) > - uses Juan's general purpose binary search instead of a custom one (it > also handles all special cases) > - uses #detectMax: instead of a specialized #inject:into: > > =============== Diff against Tools-eem.377 =============== > > Item was changed: > ----- Method: DebuggerMethodMap>>rangeForPC:contextIsActiveContext: (in > category 'source mapping') ----- > rangeForPC: contextsConcretePC contextIsActiveContext: > contextIsActiveContext > "Answer the indices in the source code for the supplied pc. > If the context is the actve context (is at the hot end of the > stack) > then its pc is the current pc. But if the context isn't, because > it is > suspended sending a message, then its current pc is the previous > pc." > > + | pc | > - | pc i end | > pc := self method abstractPCForConcretePC: (contextIsActiveContext > > ifTrue: [contextsConcretePC] > > ifFalse: [(self method pcPreviousTo: > contextsConcretePC) > > ifNotNil: [:prevpc| > prevpc] > > ifNil: > [contextsConcretePC]]). > (self abstractSourceMap includesKey: pc) ifTrue: > [^self abstractSourceMap at: pc]. > + sortedSourceMap ifNil: [ > + sortedSourceMap := self abstractSourceMap associations > + replace: [ :each | each copy ]; > + sort ]. > + ^sortedSourceMap > + findBinary: [ :assoc | pc - assoc key ] > + do: [ :item | item value ] > + ifNone: [ :lower :upper | > + lower > + ifNil: [ 1 to: 0 ] > + ifNotNil: [ > + upper > + ifNotNil: [ upper value ] > "No match, but a nearby element." > + ifNil: [ > + | end | > + end := > sortedSourceMap detectMax: [ :each | each value last ]. > + end + 1 to: end ] ] > ] > - sortedSourceMap ifNil: > - [sortedSourceMap := self abstractSourceMap. > - sortedSourceMap := (sortedSourceMap keys collect: > - [:key| key > -> (sortedSourceMap at: key)]) asSortedCollection]. > - (sortedSourceMap isNil or: [sortedSourceMap isEmpty]) ifTrue: [^1 > to: 0]. > - i := sortedSourceMap findNearbyBinaryIndex: [:assoc| pc - assoc > key]. > - i < 1 ifTrue: [^1 to: 0]. > - i > sortedSourceMap size ifTrue: > - [end := sortedSourceMap inject: 0 into: > - [:prev :this | prev max: this value last]. > - ^end+1 to: end]. > - ^(sortedSourceMap at: i) value > > "| method source scanner map | > method := DebuggerMethodMap compiledMethodAt: > #rangeForPC:contextIsActiveContext:. > source := method getSourceFromFile asString. > scanner := InstructionStream on: method. > map := method debuggerMap. > Array streamContents: > [:ranges| > [scanner atEnd] whileFalse: > [| range | > range := map rangeForPC: scanner pc > contextIsActiveContext: true. > ((map abstractSourceMap includesKey: scanner > abstractPC) > and: [range first ~= 0]) ifTrue: > [ranges nextPut: (source copyFrom: range > first to: range last)]. > scanner interpretNextInstructionFor: > InstructionClient new]]"! > > > -- best, Eliot

3 4

[CI] Generated some Jenkins/Hudson XML
by Casey Ransberger 09 Sep '11

09 Sep '11

With Levente's help (thanks for the Pharo-Squeak translation on the #run method!) I was able to get a test pass going by kicking it off in a workspace using HDTestReporter, which is used in the CI they're doing with Pharo. I just ran the kernel tests, but this is a pretty good sign that this will be fairly straightforward to make work. Not sure I like forking the code though; does anyone know if there's a way to check which fork one is in? I thought I saw something about that somewhere... another option is to use an override from another package, but methinks overrides aren't good for business. Lukas, if I can make this work for both Squeak and Pharo without problems, would you consider integrating the changes so we don't have to make a fork of it? -- Casey Ransberger

2 1

Daily Commit Log
by commits＠source.squeak.org 09 Sep '11

09 Sep '11

Changes to Trunk (http://source.squeak.org/trunk.html) in the last 24 hours: http://lists.squeakfoundation.org/pipermail/packages/2011-September/004949.… Name: Collections-eem.455 Ancestors: Collections-ul.454 Add findNearbyBinaryIndex: for DebuggerMethodMap's rangeForPC:contextIsActiveContext:. ============================================= http://lists.squeakfoundation.org/pipermail/packages/2011-September/004950.… Name: Tools-eem.377 Ancestors: Tools-bf.376 Replace use of private indexForInserting: with public findNearbyBinaryIndex: in DebuggerMethodMap>>rangeForPC:contextIsActiveContext:. N.B. Requires at least Collections-eem.455 ============================================= http://lists.squeakfoundation.org/pipermail/packages/2011-September/004951.… Name: Kernel-nice.618 Ancestors: Kernel-nice.617 Simplify Integer>>printTruncatedOn:showingDecimalPlaces: as suggested by Levente =============================================

1 0

[CI] Author, question about a Pharoism
by Casey Ransberger 08 Sep '11

08 Sep '11

I'm looking at what Lukas did for CI in Pharo, it looks wholly adaptable to Squeak AFAICT. I'm hung up on a Pharoism. I'd really like to be able to check what fork I'm in and then do the right thing, but I'll settle for forking the package (temporarily) if it means I can get a build building in short order. How would one say this in Squeak? It's the during: part that I'm not sure about; I know of Utilities>>authorInitials. Well, I suppose I'm also fuzzy on how I'd default to 'hudson' for the author... it seems like maybe I can just force the use of 'hudson' as the author and call it good... not sure I see the use case otherwise. HDTestReport>>run Author uniqueInstance ifUnknownAuthorUse: 'hudson' during: [ [ self setUp. suiteTime := [ self runAll ] timeToRun ] ensure: [ self tearDown ] ] -- Casey Ransberger

2 1

The Inbox: Tools-ul.378.mcz
by commits＠source.squeak.org 08 Sep '11

08 Sep '11

A new version of Tools was added to project The Inbox: http://source.squeak.org/inbox/Tools-ul.378.mcz ==================== Summary ==================== Name: Tools-ul.378 Author: ul Time: 8 September 2011, 10:57:10.107 pm UUID: 0262d60e-4ceb-b246-8d3f-96424d2aa1ce Ancestors: Tools-eem.377 Simplified DebuggerMethodMap >> #rangeForPC:contextIsActiveContext:. - copy and sort the associations in a bit simpler and probably faster way - removed unnecessary checks (#isNil after non-nil assignment) - uses Array instead of SortedCollection :) - uses Juan's general purpose binary search instead of a custom one (it also handles all special cases) - uses #detectMax: instead of a specialized #inject:into: =============== Diff against Tools-eem.377 =============== Item was changed: ----- Method: DebuggerMethodMap>>rangeForPC:contextIsActiveContext: (in category 'source mapping') ----- rangeForPC: contextsConcretePC contextIsActiveContext: contextIsActiveContext "Answer the indices in the source code for the supplied pc. If the context is the actve context (is at the hot end of the stack) then its pc is the current pc. But if the context isn't, because it is suspended sending a message, then its current pc is the previous pc." + | pc | - | pc i end | pc := self method abstractPCForConcretePC: (contextIsActiveContext ifTrue: [contextsConcretePC] ifFalse: [(self method pcPreviousTo: contextsConcretePC) ifNotNil: [:prevpc| prevpc] ifNil: [contextsConcretePC]]). (self abstractSourceMap includesKey: pc) ifTrue: [^self abstractSourceMap at: pc]. + sortedSourceMap ifNil: [ + sortedSourceMap := self abstractSourceMap associations + replace: [ :each | each copy ]; + sort ]. + ^sortedSourceMap + findBinary: [ :assoc | pc - assoc key ] + do: [ :item | item value ] + ifNone: [ :lower :upper | + lower + ifNil: [ 1 to: 0 ] + ifNotNil: [ + upper + ifNotNil: [ upper value ] "No match, but a nearby element." + ifNil: [ + | end | + end := sortedSourceMap detectMax: [ :each | each value last ]. + end + 1 to: end ] ] ] - sortedSourceMap ifNil: - [sortedSourceMap := self abstractSourceMap. - sortedSourceMap := (sortedSourceMap keys collect: - [:key| key -> (sortedSourceMap at: key)]) asSortedCollection]. - (sortedSourceMap isNil or: [sortedSourceMap isEmpty]) ifTrue: [^1 to: 0]. - i := sortedSourceMap findNearbyBinaryIndex: [:assoc| pc - assoc key]. - i < 1 ifTrue: [^1 to: 0]. - i > sortedSourceMap size ifTrue: - [end := sortedSourceMap inject: 0 into: - [:prev :this | prev max: this value last]. - ^end+1 to: end]. - ^(sortedSourceMap at: i) value "| method source scanner map | method := DebuggerMethodMap compiledMethodAt: #rangeForPC:contextIsActiveContext:. source := method getSourceFromFile asString. scanner := InstructionStream on: method. map := method debuggerMap. Array streamContents: [:ranges| [scanner atEnd] whileFalse: [| range | range := map rangeForPC: scanner pc contextIsActiveContext: true. ((map abstractSourceMap includesKey: scanner abstractPC) and: [range first ~= 0]) ifTrue: [ranges nextPut: (source copyFrom: range first to: range last)]. scanner interpretNextInstructionFor: InstructionClient new]]"!

1 0

[CI] Let's get together. (was Re: [squeak-dev] Re: [CI] Security)
by Casey Ransberger 08 Sep '11

08 Sep '11

Unfortunately, I have had almost no time to work on this. Unexpected family related emergency put me on finding a job sooner than I thought I'd need to. If there's anyone here who wants to race me to continuous integration for Squeak now, I will be glad to lose this race, as I really believe that this is something we need sooner than later. It would also be great to see folks collaborate on a solution; sadly, this shouldn't be too hard, but I just don't have any time. Thus my general absence from squeak-dev, etc. There are a number of CI-related packages on squeaksource, have you looked at any of these Hannes? I was investigating those when disaster struck. I found at least four different CI-related packages. I would rather not reinvent the wheel unless the wheel is quite broken, but I didn't get as far as actually trying to use these packages. Hannes, you do great work with regard to investigating various aspects of a problem and summarizing them. Would you like to help me get this part done? Most likely, and honestly, our best plan is probably to adapt the work that various awesome people have already done. This one is Igor's: http://www.squeaksource.com/CISupport This one belongs to Laurent Laffont: http://www.squeaksource.com/ContIntegration This one seems to be a collaboration between Yanni Chiu and Lukas Renggli; we might pay extra attention to this one, because for some reason I think it's what they're using for Pharo presently (though I may be wrong, don't recall why I think this) which means it's been working in production use: http://www.squeaksource.com/HudsonBuild And last but not least, here's another one worth looking at, a la Andreas Raab: http://www.squeaksource.com/InT My plan was to take a survey of these, contact the maintainers, weigh the pros and cons, and report back to the list. If there was consensus around one of them, I'd run with that, if there wasn't, I would reinvent the wheel. Thanks for lighting a fire up under my ___, Hannes :D Everyday life so often distracts from the things that really matter, like Squeak! On Tue, Sep 6, 2011 at 2:00 AM, H. Hirzel <hannes.hirzel(a)gmail.com> wrote: > Are there any news about setting up CI? > > --Hannes <biiiiig snip>

2 2

Re: [squeak-dev] The Trunk: Kernel-nice.618.mcz
by Levente Uzonyi 08 Sep '11

08 Sep '11

On Thu, 8 Sep 2011, commits(a)source.squeak.org wrote: > Nicolas Cellier uploaded a new version of Kernel to project The Trunk: > http://source.squeak.org/trunk/Kernel-nice.618.mcz > > ==================== Summary ==================== > > Name: Kernel-nice.618 > Author: nice > Time: 8 September 2011, 10:26:12.614 pm > UUID: 5c9061ca-b336-460a-bd15-2dfb6ecd2d0c > Ancestors: Kernel-nice.617 > > Simplify Integer>>printTruncatedOn:showingDecimalPlaces: as suggested by Levente I also prepared this change as you suggested, but got distracted by Eliot's changes and some bugs in a production system... Thanks, Levente > > =============== Diff against Kernel-nice.617 =============== > > Item was changed: > ----- Method: Integer>>printTruncatedOn:showingDecimalPlaces: (in category 'printing') ----- > printTruncatedOn: aStream showingDecimalPlaces: placesDesired > "Print a representation of the receiver on aStream in decimal notation with prescribed number of places after decimal separator. > Print as if the receiver was truncated to requested precision." > > self printOn: aStream. > placesDesired > 0 > ifTrue: > [aStream nextPut: $.. > + aStream next: placesDesired put: $0]! > - aStream nextPutAll: (String new: placesDesired withAll: (Character digitValue: 0))]! > > >

1 0

The Trunk: Kernel-nice.618.mcz
by commits＠source.squeak.org 08 Sep '11

08 Sep '11

Nicolas Cellier uploaded a new version of Kernel to project The Trunk: http://source.squeak.org/trunk/Kernel-nice.618.mcz ==================== Summary ==================== Name: Kernel-nice.618 Author: nice Time: 8 September 2011, 10:26:12.614 pm UUID: 5c9061ca-b336-460a-bd15-2dfb6ecd2d0c Ancestors: Kernel-nice.617 Simplify Integer>>printTruncatedOn:showingDecimalPlaces: as suggested by Levente =============== Diff against Kernel-nice.617 =============== Item was changed: ----- Method: Integer>>printTruncatedOn:showingDecimalPlaces: (in category 'printing') ----- printTruncatedOn: aStream showingDecimalPlaces: placesDesired "Print a representation of the receiver on aStream in decimal notation with prescribed number of places after decimal separator. Print as if the receiver was truncated to requested precision." self printOn: aStream. placesDesired > 0 ifTrue: [aStream nextPut: $.. + aStream next: placesDesired put: $0]! - aStream nextPutAll: (String new: placesDesired withAll: (Character digitValue: 0))]!

1 0

The Trunk: Tools-eem.377.mcz
by commits＠source.squeak.org 08 Sep '11

08 Sep '11

Eliot Miranda uploaded a new version of Tools to project The Trunk: http://source.squeak.org/trunk/Tools-eem.377.mcz ==================== Summary ==================== Name: Tools-eem.377 Author: eem Time: 8 September 2011, 10:46:07.752 am UUID: 620788e4-8d15-41bd-af90-58b26f705630 Ancestors: Tools-bf.376 Replace use of private indexForInserting: with public findNearbyBinaryIndex: in DebuggerMethodMap>>rangeForPC:contextIsActiveContext:. N.B. Requires at least Collections-eem.455 =============== Diff against Tools-bf.376 =============== Item was changed: ----- Method: DebuggerMethodMap>>rangeForPC:contextIsActiveContext: (in category 'source mapping') ----- rangeForPC: contextsConcretePC contextIsActiveContext: contextIsActiveContext "Answer the indices in the source code for the supplied pc. If the context is the actve context (is at the hot end of the stack) then its pc is the current pc. But if the context isn't, because it is suspended sending a message, then its current pc is the previous pc." | pc i end | pc := self method abstractPCForConcretePC: (contextIsActiveContext ifTrue: [contextsConcretePC] ifFalse: [(self method pcPreviousTo: contextsConcretePC) ifNotNil: [:prevpc| prevpc] ifNil: [contextsConcretePC]]). (self abstractSourceMap includesKey: pc) ifTrue: [^self abstractSourceMap at: pc]. sortedSourceMap ifNil: [sortedSourceMap := self abstractSourceMap. sortedSourceMap := (sortedSourceMap keys collect: [:key| key -> (sortedSourceMap at: key)]) asSortedCollection]. (sortedSourceMap isNil or: [sortedSourceMap isEmpty]) ifTrue: [^1 to: 0]. + i := sortedSourceMap findNearbyBinaryIndex: [:assoc| pc - assoc key]. - i := sortedSourceMap indexForInserting: (pc -> nil). i < 1 ifTrue: [^1 to: 0]. i > sortedSourceMap size ifTrue: [end := sortedSourceMap inject: 0 into: [:prev :this | prev max: this value last]. ^end+1 to: end]. ^(sortedSourceMap at: i) value "| method source scanner map | method := DebuggerMethodMap compiledMethodAt: #rangeForPC:contextIsActiveContext:. source := method getSourceFromFile asString. scanner := InstructionStream on: method. map := method debuggerMap. Array streamContents: [:ranges| [scanner atEnd] whileFalse: [| range | range := map rangeForPC: scanner pc contextIsActiveContext: true. ((map abstractSourceMap includesKey: scanner abstractPC) and: [range first ~= 0]) ifTrue: [ranges nextPut: (source copyFrom: range first to: range last)]. scanner interpretNextInstructionFor: InstructionClient new]]"!

1 0

The Trunk: Collections-eem.455.mcz
by commits＠source.squeak.org 08 Sep '11

08 Sep '11

Eliot Miranda uploaded a new version of Collections to project The Trunk: http://source.squeak.org/trunk/Collections-eem.455.mcz ==================== Summary ==================== Name: Collections-eem.455 Author: eem Time: 8 September 2011, 10:43:02.64 am UUID: 717b6721-47e5-4edb-aaa9-66e942c4943a Ancestors: Collections-ul.454 Add findNearbyBinaryIndex: for DebuggerMethodMap's rangeForPC:contextIsActiveContext:. =============== Diff against Collections-ul.454 =============== Item was added: + ----- Method: SequenceableCollection>>findNearbyBinaryIndex: (in category 'enumerating') ----- + findNearbyBinaryIndex: aBlock + "Search for an element in the receiver using binary search. + The argument aBlock is a one-element block returning + 0 - if the element is the one searched for + <0 - if the search should continue in the first half + >0 - if the search should continue in the second half + If no matching element is found, answer the closest index we could find, + answering 0 if the element should preceed all items in the collection, + and size + 1 if the element should follow all items in the collection." + | index low high test | + low := 1. + high := self size. + [index := high + low // 2. + low > high] whileFalse:[ + test := aBlock value: (self at: index). + test = 0 + ifTrue:[^index] + ifFalse:[test > 0 + ifTrue: [low := index + 1] + ifFalse: [high := index - 1]]]. + ^index = self size ifTrue: [index + 1] ifFalse: [index]!

1 0

Jump to page: