- Pharo-dev - lists.pharo.org

Re: [Pharo-project] [squeak-dev] Xtreams up to date

by mkobetic＠gmail.com

"Hannes Hirzel"<hannes.hirzel(a)gmail.com> wrote: > If I explore the resulting object I got what is shown in the screen > shot. I am not sure how to interpret the result. Yes, it's not the best possible representation, but it does show the full parse tree. The tree can be rather complicated depending on how complicated is the grammar. That's why I suggested playing with simpler grammars. But the reason why I even mentioned the TreeBuilder is that it's the simplest possible Actor, it implements single method that packages the result of each successful production as it's own node. Maybe using an array instead of associations would create a slightly more "tree like" result. > "Regarding the wiki example: the class PEGWikiGenerator is not included" > > wikiParser parse: 'Page' stream: input actor: PEGWikiGenerator new. > > Where can I get this class? I've attached a fileout of the class that you can file into Squeak, except it references VW XML classes, so it won't work. If you fix it up to use the Squeak XML stuff you should be able to get it to work. All you really need is XML.Element and XML.Tag equivalent. The second fileout shows how the class variables are initialized in VW. You might need to turn the initializers into equivalent class side initialize methods. > As class methods of PEGParser there are 8 grammars included - CSS3, > HttpUrl, JSON, JavaScript, PEG, Smalltalk, Wiki, XML. > > I think a simple complete example which actually does something would > be helpful for me. TreeBuilder seems to do nothing and I do not see the tree. OK, let's try something really simple. The wikipedia page shows a simple arithmetic expression grammar. Here it is transcribed for Xtreams: grammar := ' Number <- [0-9]+ Parenthesised <- "(" Expr ")" Value <- Number / Parenthesised Product <- Value (("*" / "/") Value)* Sum <- Product (("+" / "-") Product)* Expr <- Sum'. I expanded the Value from the wiki grammar into separate Number and Parethesised expression productions. It'll hopefully be clear why shortly. You already know how to get a parser from grammar. That's always the same and for anything more complex you'll probably want to cache the parser instead of recreating it every time, but this will do fine for now: parser := PEG.Parser parserPEG parse: 'Grammar' stream: grammar actor: PEG.ParserParser new. With this you can already parse expressions (note that the grammar as it is written cannot handle whitespace) parser parse: 'Expr' stream: '3+4*(10-4)' reading actor: nil you'll get this (roughly formatted to provide at least some resemblance of the expression tree): #( #(OrderedCollection ($3 "16r0033") OrderedCollection ()) OrderedCollection (# ('+' #(OrderedCollection ($4 "16r0034") OrderedCollection (# ('*' #( '(' #( #(OrderedCollection ($1 "16r0031" $0 "16r0030") OrderedCollection ()) OrderedCollection (# ('-' #(OrderedCollection ($4 "16r0034") OrderedCollection ())))) ')'))))))) That's obviously not very easy to take apart to work with, that's why you want an Actor that will receive rule specific callbacks and do whatever. Let's say we want to evaluate the expression and return the result. We'll make a subclass of Actor called Evaluator. As I mentioned earlier, the Actor class provides pragma based mechanism for expressing rule specific callbacks so you simply write a method for each rule for which you want to do something and annotate it with a pragma. Let's first try to turn the digit collections into numbers: Evaluator>>Number: digits <action: 'Number'> ^digits inject: 0 into: [ :total :digit | total * 10 + ('0123456789' indexOf: digit) - 1 ] With this you can parse numbers parser parse: 'Expr' stream: '3456' reading actor: Evaluator new Now let's try to add. We'll want a method for the Sum production but there's a slightly tricky part there. Note that the rule is defined as Sum <- Product (("+" / "-") Product)* which means that you get a term (Product) and then zero or more pairs of + or - and another term. What the method receives will be collection of the first term and then pairs of the operator with its term. So parsing this: parser parse: 'Expr' stream: '3+4+5+6+7' reading actor: Evaluator new will give you #(3 OrderedCollection (#('+' 4) #('+' 5) #('+' 6) #('+' 7))) So if you add a method for Sum, that's what it will need to deal with: Sum: terms <action: 'Sum'> ^terms last inject: terms first into: [ :total :next | (next first = '+') ifTrue: [ total + next last ] ifFalse: [ total - next last ] ] The Actor pragmas provide another convenience for taking more complex input apart. You can write the above as Sum: first rest: pairs <action: 'Sum' arguments: #(1 2)> ^pairs inject: first into: [ :total :pair | (pair first = '+') ifTrue: [ total + pair last ] ifFalse: [ total - pair last ] ] The arguments: keyword simply says invoke this method with elements 1 and 2 of the rule input (terms) as arguments. Obviously the method must take the corresponding number of arguments as well. So with the above rules the addition expression should give you 25. The Product rule handling is analogous: Product: first rest: pairs <action: 'Product' arguments: #(1 2)> ^pairs inject: first into: [ :total :pair | (pair first = '*') ifTrue: [ total * pair last ] ifFalse: [ total / pair last ] ] Finally the parenthesised expression is rather simple: Parenthesised: expression <action: 'Parenthesised' arguments: #(2)> ^expression It receives a three element array: the left bracket, the expression inside and the right bracket. All we really need to do is return the expression in the middle. The above is using the arguments: pragma again to achieve that, the following would be equivalent: Parenthesised: expression <action: 'Parenthesised'> ^expression at: 2 I hope by now it's also clearer why it was useful to factor the Value definition out into separate Number and Parenthesised rule. Otherwise we'd have to handle both cases in a single Value: method, which would make the code messier. > For example besides the classes PEGParserPEGTest and > PEGParserSmalltalkTest (6 test cases) a class PEGParserHttpUrlTest or > PEGParserJavaScriptTest would be helpful. Yup, the tests are seriously wanting as well. Hopefully we'll get there soon. BTW, the parsing stuff doesn't work for me out of the box in either Squeak or Pharo in the images that I use (which should be quite clean). The problem I'm seeing is that the Dictionary>>addAll: in ParserParser>>Grammar: doesn't work as expected. It's trying to add collection of associations to a dictionary but ends up adding them as values indexed by integers. I can fix it but I wonder what do people have loaded that makes it work for them. Anyway, hopefully the above can get you going. Cheers, Martin

Jan. 23, 2011

[Pharo-project] closeable tabs

by Tudor Girba

Hi, I worked a bit on adding the possibility of closing tabs from a tab group. This is available for the moment in the I refactored the LazyTabGroupMorph from the Glamour repository. It would still be good to delegate a couple of things to the Theme, but other than that I think we could replace the logic from TabGroupMorph. What I did was to introduce a LazyTabPage that can lazily draw the page morph. If you (Alain, Gary?) have time, I could use a code review :). Gofer new squeaksource: 'Glamour'; package: 'ConfigurationOfGlamour'; load. (Smalltalk at: #ConfigurationOfGlamour) perform: #loadDefault Cheers, Doru -- www.tudorgirba.com "In a world where everything is moving ever faster, one might have better chances to win by moving slower."

Jan. 23, 2011

Re: [Pharo-project] IMO IllegalResumeAttempt should be resumable

by Eliot Miranda

Belay that. Apologies for the noise. resumeUnchecked: is of course the right solution. e.g. Array streamContents: [:s| [Cogit chooseCogitClass testPCMappingSelect: [:m| true]] on: AssertionFailure, Error do: [:ex| | ctxt | ctxt := ex signalerContext findContextSuchThat: [:ctx| ctx sender selector = #allSelect:]. "testPCMappingSelect uses allSelect: to enumerate so return to allSelect: to continue" s nextPut: (ctxt tempAt: 1). Transcript ensureCr; cr; print: (ctxt tempAt: 1); cr; flush. ex searchFrom: ctxt. ex resumeUnchecked: false]] On Sat, Jan 22, 2011 at 1:50 PM, Eliot Miranda <eliot.miranda(a)gmail.com>wrote: > Hi All, > > an attempt to resume an unresumable signal results in > raising IllegalResumeAttempt; quite right: > > Exception>>resume > "Return from the message that signaled the receiver." > > self resume: self defaultResumeValue > > > However, IllegalResumeAttempt is not itself resumable. So in those rare > cases (e.g. test coverage) where a knowledgeable client wants to force > resumption of a non-resumable signal by catching and resuming > IllegalResumeAttempt, they can't because > > IllegalResumeAttempt>>isResumable > ^ false > > > So take e.g. > > [Cogit chooseCogitClass testPCMappingSelect: [:m| true]] > on: AssertionFailure > do: [:ex| | ctxt | > ctxt := ex signalerContext findContextSuchThat: > [:ctx| ctx sender selector = #allSelect:]. > Transcript ensureCr; print: (ctxt tempAt: 1); flush. > [ex resume] > on: IllegalResumeAttempt > do: [:ex| ex resume]] > > where I'm trying to JIT every method in my image and catalogue those that > assert-fail, proceeding through the assert failure (and AssertionFailure > isn't resumable). I happen to know that proceeding from the assert-fails is > safe. But I can't force resumption because IllegalResumeAttempt is itself > not resumable. > > I propose that we make IllegalResumeAttempt resumable. Objections? > Cautions? > > best > Eliot >

Jan. 22, 2011

Re: [Pharo-project] Some important news

by GermÃ¡n Arduino

+1 2011/1/22 Esteban Lorenzano <estebanlm(a)gmail.com>: > +1 > > El 21/01/2011, a las 11:47p.m., Tony Fleig escribiÃ³: > >> Excellent. Where do I sign up? >> >> On Fri, Jan 21, 2011 at 11:26 AM, StÃ©phane Ducasse >> <stephane.ducasse(a)inria.fr> wrote: >>> Hi guys >>> >>> some important news today, we decided that we will set up a pharo association >>> so that we can promote pharo. We would like to be able to collect money and >>> create bounties to improve pharo. >>> >>> Stef >>> >> > > > -- ================================================= GermÃ¡n S. ArduinoÂ <gsa @ arsol.net>Â Â Twitter: garduino Arduino Software & Web HostingÂ Â http://www.arduinosoftware.com PasswordsProÂ http://www.passwordspro.com =================================================

Jan. 22, 2011

[Pharo-project] IMO IllegalResumeAttempt should be resumable

by Eliot Miranda

Hi All, an attempt to resume an unresumable signal results in raising IllegalResumeAttempt; quite right: Exception>>resume "Return from the message that signaled the receiver." self resume: self defaultResumeValue However, IllegalResumeAttempt is not itself resumable. So in those rare cases (e.g. test coverage) where a knowledgeable client wants to force resumption of a non-resumable signal by catching and resuming IllegalResumeAttempt, they can't because IllegalResumeAttempt>>isResumable ^ false So take e.g. [Cogit chooseCogitClass testPCMappingSelect: [:m| true]] on: AssertionFailure do: [:ex| | ctxt | ctxt := ex signalerContext findContextSuchThat: [:ctx| ctx sender selector = #allSelect:]. Transcript ensureCr; print: (ctxt tempAt: 1); flush. [ex resume] on: IllegalResumeAttempt do: [:ex| ex resume]] where I'm trying to JIT every method in my image and catalogue those that assert-fail, proceeding through the assert failure (and AssertionFailure isn't resumable). I happen to know that proceeding from the assert-fails is safe. But I can't force resumption because IllegalResumeAttempt is itself not resumable. I propose that we make IllegalResumeAttempt resumable. Objections? Cautions? best Eliot

Jan. 22, 2011

Re: [Pharo-project] Issue 3465 in pharo: Hundson OneClick: Does not show Mac menues

by pharo＠googlecode.com

Updates: Labels: -Milestone-1.2-DevImage Milestone-1.3 Milestone-1.3 Comment #3 on issue 3465 by marianopeck: Hundson OneClick: Does not show Mac menues http://code.google.com/p/pharo/issues/detail?id=3465 I move this to Pharo 1.3

Jan. 22, 2011

Re: [Pharo-project] Issue 3578 in pharo: Hudson integration: Not include MAC files in zips

by pharo＠googlecode.com

Updates: Status: Closed Comment #1 on issue 3578 by marcus.d...(a)gmail.com: Hudson integration: Not include MAC files in zips http://code.google.com/p/pharo/issues/detail?id=3578 Yes, I will do that.

Jan. 22, 2011

Re: [Pharo-project] Issue 3578 in pharo: Hudson integration: Not include MAC files in zips

by pharo＠googlecode.com

Comment #2 on issue 3578 by marcus.d...(a)gmail.com: Hudson integration: Not include MAC files in zips http://code.google.com/p/pharo/issues/detail?id=3578 Yes, I will do that.

Jan. 22, 2011

[Pharo-project] Issue 3578 in pharo: Hudson integration: Not include MAC files in zips

by pharo＠googlecode.com

Status: Accepted Owner: marianopeck CC: marcus.d...(a)gmail.com, siguc...(a)gmail.com Labels: Milestone-1.3 New issue 3578 by marianopeck: Hudson integration: Not include MAC files in zips http://code.google.com/p/pharo/issues/detail?id=3578 Igor: All .zip files with pharo images having an annoying Apple-ism - a .DS_Store garbage file. Please use a zip archiver which includes only relevant files to archive, because obviously this garbage has no use anywhere but Macs. Mariano: Yes, when I build the dev images, I use the YemuZip that Marcus recommended. It can zip it without all that crap. Not sure if it is the best way if you want to automatize this with scrips. We need to integrate this for hudson. cheers mariano

Jan. 22, 2011

Re: [Pharo-project] Reminder: change the archiving tool

by Mariano Martinez Peck

Yes, when I build the dev images, I use the YemuZip that Marcus recommended. It can zip it without all that crap. Not sure if it is the best way if you want to automatize this with scrips. http://code.google.com/p/pharo/issues/detail?id=3578 cheers mariano On Fri, Jan 21, 2011 at 5:19 PM, Miguel CobÃ¡ <miguel.coba(a)gmail.com> wrote: > El vie, 21-01-2011 a las 13:59 +0100, Igor Stasenko escribiÃ³: > > All .zip files with pharo images having an annoying Apple-ism - a > > .DS_Store garbage file. > > Please use a zip archiver which includes only relevant files to > > archive, because obviously this garbage has no use anywhere but Macs. > > > > +1 > > -- > Miguel CobÃ¡ > http://twitter.com/MiguelCobaMtz > http://miguel.leugim.com.mx > > > > >

Jan. 22, 2011