About validating the result of RBParser >> parseMethod:
Locked
About validating the result of RBParser >> parseMethod:
 
|
I have a question about validating a methodNode, suppose we have the following:
parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. Is there a way of checking the health/validity of the parsedNode? I couldn't find any behavior in RBMethodNode or related classes. Lukas, do you have an idea is i could run any Lint rule? My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. Thanks, Fernando _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
You can use #parseMethod:onError: to check for syntactical errors.
Semantical errors are harder to detect, some of them (e.g. undeclared variables) you can find using AST-Semantics. Lukas On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: > I have a question about validating a methodNode, suppose we have the following: > > parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. > > Is there a way of checking the health/validity of the parsedNode? > > I couldn't find any behavior in RBMethodNode or related classes. > > Lukas, do you have an idea is i could run any Lint rule? > > My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. > > Thanks, > Fernando > > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- Lukas Renggli www.lukas-renggli.ch _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
Lukas, thanks for the answer. I will see if AST-Semantics solves my problem.
I also tried using the Compiler and the parser, and faking a notifier. But the Compiler>>interactive behavior is hard to override, it always gives me visual feedback ( pop menus and related dialogs). Maybe the newCompiler is more customizable? Fernando On Jul 23, 2010, at 7:34 AM, Lukas Renggli wrote: > You can use #parseMethod:onError: to check for syntactical errors. > > Semantical errors are harder to detect, some of them (e.g. undeclared > variables) you can find using AST-Semantics. > > Lukas > > On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: >> I have a question about validating a methodNode, suppose we have the following: >> >> parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. >> >> Is there a way of checking the health/validity of the parsedNode? >> >> I couldn't find any behavior in RBMethodNode or related classes. >> >> Lukas, do you have an idea is i could run any Lint rule? >> >> My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. >> >> Thanks, >> Fernando >> >> >> _______________________________________________ >> Pharo-project mailing list >> [hidden email] >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> > > > > -- > Lukas Renggli > www.lukas-renggli.ch > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
Thanks for the pointers Lukas,
Works now! >>canSuccesfullyParse: aSource in: aClass [ parsedNode := RBParser parseMethod: aSource onError: [:descrip :position| evaluation := false. errorDescription := descrip. errorPosition := position ] ] on: Error do:[:err| ]. parsedNode annotateInClass: aClass. hasUndeclared := parsedNode allChildren select:[:each| each isVariable and:[ each variableBinding isUndeclaredBinding ]] . hasUndeclared ifTrue:[ self .... ]. On Jul 23, 2010, at 11:46 AM, Fernando olivero wrote: > Lukas, thanks for the answer. I will see if AST-Semantics solves my problem. > > I also tried using the Compiler and the parser, and faking a notifier. But the Compiler>>interactive behavior is hard to override, > it always gives me visual feedback ( pop menus and related dialogs). Maybe the newCompiler is more customizable? > > Fernando > > > On Jul 23, 2010, at 7:34 AM, Lukas Renggli wrote: > >> You can use #parseMethod:onError: to check for syntactical errors. >> >> Semantical errors are harder to detect, some of them (e.g. undeclared >> variables) you can find using AST-Semantics. >> >> Lukas >> >> On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: >>> I have a question about validating a methodNode, suppose we have the following: >>> >>> parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. >>> >>> Is there a way of checking the health/validity of the parsedNode? >>> >>> I couldn't find any behavior in RBMethodNode or related classes. >>> >>> Lukas, do you have an idea is i could run any Lint rule? >>> >>> My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. >>> >>> Thanks, >>> Fernando >>> >>> >>> _______________________________________________ >>> Pharo-project mailing list >>> [hidden email] >>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >>> >> >> >> >> -- >> Lukas Renggli >> www.lukas-renggli.ch >> >> _______________________________________________ >> Pharo-project mailing list >> [hidden email] >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
In reply to this post by Fernando olivero
Unfortunately the NewCompiler doesn't work yet.
On 23 July 2010 11:46, Fernando olivero <[hidden email]> wrote: > Lukas, thanks for the answer. I will see if AST-Semantics solves my problem. > > I also tried using the Compiler and the parser, and faking a notifier. But the Compiler>>interactive behavior is hard to override, > it always gives me visual feedback ( pop menus and related dialogs). Maybe the newCompiler is more customizable? > > Fernando > > > On Jul 23, 2010, at 7:34 AM, Lukas Renggli wrote: > >> You can use #parseMethod:onError: to check for syntactical errors. >> >> Semantical errors are harder to detect, some of them (e.g. undeclared >> variables) you can find using AST-Semantics. >> >> Lukas >> >> On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: >>> I have a question about validating a methodNode, suppose we have the following: >>> >>> parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. >>> >>> Is there a way of checking the health/validity of the parsedNode? >>> >>> I couldn't find any behavior in RBMethodNode or related classes. >>> >>> Lukas, do you have an idea is i could run any Lint rule? >>> >>> My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. >>> >>> Thanks, >>> Fernando >>> >>> >>> _______________________________________________ >>> Pharo-project mailing list >>> [hidden email] >>> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >>> >> >> >> >> -- >> Lukas Renggli >> www.lukas-renggli.ch >> >> _______________________________________________ >> Pharo-project mailing list >> [hidden email] >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- Lukas Renggli www.lukas-renggli.ch _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
In reply to this post by Fernando olivero
On 23 July 2010 11:58, Fernando olivero <[hidden email]> wrote:
> Thanks for the pointers Lukas, > > Works now! > >>>canSuccesfullyParse: aSource in: aClass > [ parsedNode := RBParser parseMethod: aSource onError: [:descrip :position| evaluation := false. errorDescription := descrip. errorPosition := position ] ] on: Error do:[:err| ]. > parsedNode annotateInClass: aClass. > hasUndeclared := parsedNode allChildren select:[:each| each isVariable and:[ each variableBinding isUndeclaredBinding ]] . > hasUndeclared ifTrue:[ self .... ]. Yep, something like that. You shouldn't need the Error handler, otherwise that would be a bug in the RBParser. Also you might want to check for shadowed variables (#isShadowed), because they cause the compiler to complain. Lukas -- Lukas Renggli www.lukas-renggli.ch _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
In reply to this post by Fernando olivero
Hi Fernando,
On Fri, Jul 23, 2010 at 2:46 AM, Fernando olivero <[hidden email]> wrote: Lukas, thanks for the answer. I will see if AST-Semantics solves my problem. Colin Putney has done good work replacing a number of interactive errors with exceptions whose defaultAction is to produce the notifier. These are in Squeak 4.1. Have you taken a look at the trunk 4.1 compiler? Clearly the standard compiler should be able to do what you want. We should fix it :)
best Eliot Maybe the newCompiler is more customizable? _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
I thought I integrated those.
Fernando if you can check and let us know. Stef On Jul 23, 2010, at 6:48 PM, Eliot Miranda wrote: > Hi Fernando, > > On Fri, Jul 23, 2010 at 2:46 AM, Fernando olivero <[hidden email]> wrote: > Lukas, thanks for the answer. I will see if AST-Semantics solves my problem. > > I also tried using the Compiler and the parser, and faking a notifier. But the Compiler>>interactive behavior is hard to override, > it always gives me visual feedback ( pop menus and related dialogs). > > Colin Putney has done good work replacing a number of interactive errors with exceptions whose defaultAction is to produce the notifier. These are in Squeak 4.1. Have you taken a look at the trunk 4.1 compiler? Clearly the standard compiler should be able to do what you want. We should fix it :) > > best > Eliot > > Maybe the newCompiler is more customizable? > > Fernando > > > On Jul 23, 2010, at 7:34 AM, Lukas Renggli wrote: > > > You can use #parseMethod:onError: to check for syntactical errors. > > > > Semantical errors are harder to detect, some of them (e.g. undeclared > > variables) you can find using AST-Semantics. > > > > Lukas > > > > On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: > >> I have a question about validating a methodNode, suppose we have the following: > >> > >> parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. > >> > >> Is there a way of checking the health/validity of the parsedNode? > >> > >> I couldn't find any behavior in RBMethodNode or related classes. > >> > >> Lukas, do you have an idea is i could run any Lint rule? > >> > >> My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. > >> > >> Thanks, > >> Fernando > >> > >> > >> _______________________________________________ > >> Pharo-project mailing list > >> [hidden email] > >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > >> > > > > > > > > -- > > Lukas Renggli > > www.lukas-renggli.ch > > > > _______________________________________________ > > Pharo-project mailing list > > [hidden email] > > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
Hi Stéphane,
On Fri, Jul 23, 2010 at 10:12 AM, Stéphane Ducasse <[hidden email]> wrote: I thought I integrated those. My point was that while good work on non-interactive errors in the standard compiler has been done there's no guarantee its finished and we should check and make whatever modifications are required so that it all errors can be caught.
best Eliot Fernando if you can check and let us know. _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
In reply to this post by Stéphane Ducasse
Yeah, these notifications are integrated in Pharo 1.0. OmniBrowser uses them.
Lukas On 23 July 2010 19:12, Stéphane Ducasse <[hidden email]> wrote: > I thought I integrated those. > Fernando if you can check and let us know. > > Stef > > On Jul 23, 2010, at 6:48 PM, Eliot Miranda wrote: > >> Hi Fernando, >> >> On Fri, Jul 23, 2010 at 2:46 AM, Fernando olivero <[hidden email]> wrote: >> Lukas, thanks for the answer. I will see if AST-Semantics solves my problem. >> >> I also tried using the Compiler and the parser, and faking a notifier. But the Compiler>>interactive behavior is hard to override, >> it always gives me visual feedback ( pop menus and related dialogs). >> >> Colin Putney has done good work replacing a number of interactive errors with exceptions whose defaultAction is to produce the notifier. These are in Squeak 4.1. Have you taken a look at the trunk 4.1 compiler? Clearly the standard compiler should be able to do what you want. We should fix it :) >> >> best >> Eliot >> >> Maybe the newCompiler is more customizable? >> >> Fernando >> >> >> On Jul 23, 2010, at 7:34 AM, Lukas Renggli wrote: >> >> > You can use #parseMethod:onError: to check for syntactical errors. >> > >> > Semantical errors are harder to detect, some of them (e.g. undeclared >> > variables) you can find using AST-Semantics. >> > >> > Lukas >> > >> > On 23 July 2010 00:35, Fernando olivero <[hidden email]> wrote: >> >> I have a question about validating a methodNode, suppose we have the following: >> >> >> >> parsedNode := RBParser parseMethod: 'test ^ self add: anUnbindedVariable'. >> >> >> >> Is there a way of checking the health/validity of the parsedNode? >> >> >> >> I couldn't find any behavior in RBMethodNode or related classes. >> >> >> >> Lukas, do you have an idea is i could run any Lint rule? >> >> >> >> My idea is to provide a method editor morph, that prevents from accepting invalid sources, and informs the user any problems detected with the edited code. >> >> >> >> Thanks, >> >> Fernando >> >> >> >> >> >> _______________________________________________ >> >> Pharo-project mailing list >> >> [hidden email] >> >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> >> >> > >> > >> > >> > -- >> > Lukas Renggli >> > www.lukas-renggli.ch >> > >> > _______________________________________________ >> > Pharo-project mailing list >> > [hidden email] >> > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> >> >> _______________________________________________ >> Pharo-project mailing list >> [hidden email] >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project >> >> _______________________________________________ >> Pharo-project mailing list >> [hidden email] >> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > > > _______________________________________________ > Pharo-project mailing list > [hidden email] > http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project > -- Lukas Renggli www.lukas-renggli.ch _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
Re: About validating the result of RBParser >> parseMethod:
|
|
In reply to this post by Eliot Miranda-2
Ok ;)
> My point was that while good work on non-interactive errors in the standard compiler has been done there's no guarantee its finished and we should check and make whatever modifications are required so that it all errors can be caught. > > best > Eliot > > Fernando if you can check and let us know. _______________________________________________ Pharo-project mailing list [hidden email] http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project |
«
Return to Pharo Smalltalk Developers
|
1 view|%1 views
| Free forum by Nabble | Edit this page |