- Pharo-users - lists.pharo.org

Re: [Pharo-users] MetaLinks with arbitrary number of arguments

by denker

> On 20 Mar 2017, at 16:34, denker <marcus.denker(a)inria.fr> wrote: > > Hi, > > Here is a slice that implements the #argsAsArray option: > > https://pharo.fogbugz.com/f/cases/19857/add-option-argsAsArray > > from a speed perspective, the question is if a dictionary makes senseâ¦ as it would > need to be created at runtime. > This test show how to use it: testReifySendArgsAsArray | sendNode instance| sendNode := (ReflectivityExamples>>#exampleMethod) ast body statements first value. link := MetaLink new metaObject: self; selector: #tagExec:; arguments: #(selector context); options: #(argsAsArray). sendNode link: link. self assert: sendNode hasMetalink. self assert: (ReflectivityExamples>>#exampleMethod) class = ReflectiveMethod. self assert: (tag isNil). instance := ReflectivityExamples new . self assert: (instance exampleMethod = 5). self assert: (tag isArray). self assert: (tag first == #+). self assert: (tag second class == Context).

March 20, 2017

Re: [Pharo-users] MetaLinks with arbitrary number of arguments

by denker

Hi, Here is a slice that implements the #argsAsArray option: https://pharo.fogbugz.com/f/cases/19857/add-option-argsAsArray from a speed perspective, the question is if a dictionary makes senseâ¦ as it would need to be created at runtime. Marcus > On 15 Mar 2017, at 11:25, denker <marcus.denker(a)inria.fr> wrote: > > >> On 15 Mar 2017, at 11:09, Peter Uhnak <i.uhnak(a)gmail.com> wrote: >> >> On Wed, Mar 15, 2017 at 10:49:51AM +0100, denker wrote: >>> >>>> On 15 Mar 2017, at 10:35, Peter Uhnak <i.uhnak(a)gmail.com> wrote: >>>> >>>> Hi, >>>> >>>> is it possible to create a metalink on arbitrary amount of arguments? >>>> >>>> Right now the metaObject and selector must have the same number of arguments, e.g. >>>> >>>> MetaLink new >>>> metaObject: [ :object :selector :args | ... ]; >>>> selector: #value:value:value; >>>> arguments: #(object selector arguments) >>>> >>>> >>>> What I would like to have instead is something like... >>>> >>>> MetaLink new >>>> metaObject: [ :allArgs | ... ]; >>>> selector: #value:; >>>> arguments: #(object selector arguments) >>>> >>>> >>>> And then `allArgs` would either be an array with all the arguments, or maybe a Dictionary. >>>> >>> >>> Right now notâ¦ but it would be a nice addition. >>> >>> It could either use some magic (if the number of args requested are larger then the number of args >>> of the selector it gives the rest as a dictionary). >>> >>> or, maybe better (less magic), this could be an option to be configured per link: >>> >>> >>> MetaLink new >>> metaObject: [ :allArgs | ... ]; >>> selector: #value:; >>> arguments: #(object selector arguments) >>> options: #(argsAsDictionary) >>> >>> (which would allow #argsAsArray, too). >>> >>> Marcus >> >> >> +1 for option, magic only brings suffering. :) >> > > This should be easy.. I will look into it hopefully this week. > > Marcus >

March 20, 2017

read, display, and annotate MS Word and PDF files?

by David Epstein

Hi folks. I am considering trying Pharo for a project but am unfamiliar with Smalltalk. I'm attracted by the close integration between language and IDE. Are there native libraries that would allow me to read, display, and annotate (e.g. add comments & highlighting) to MS Word and PDF files? If not, is it easy to call libraries intended for other languages, such as http://www.foxitsdk.com/ ? Thank you, -david

March 20, 2017

Re: [Pharo-users] Magritte extension

by Esteban A. Maringolo

2017-03-20 7:18 GMT-03:00 Otto Behrens <otto(a)finworks.biz>: > Hi, > > We also made several sub-classes of Magritte classes and / or > components. This causes a lot of problems because when we upgraded > Magritte it was difficult. It also shows that Magritte (the core > classes) cannot change much because the impact will be high for most > users of Magritte. This makes us think that Magritte is not easy to > use or customise. And it also appears as if all the custom changes are > not being generalised into the framework. Perhaps us users of Magritte > should stand back and discuss our changes and issues more so that we > can improve it? I suppose we need some magritte champion(s) to take it > somewhere. > > To get more flexibility with Magritte styling we are using CSS and > classes more in stead of rendering in a different way. Often one can > get away with adding a CSS class and changing the style sheet. > > This is a bit fluffy, but it there's a lot of stuff to talk about > here. In principle, are you sharing my frustration with Magritte? I'm not frustrated, but I share the feeling that we're missing valuable contributions from other users. The thing with Magritte is that everything is done by subclassing, the metamodel part (MADescription subclasses) is pretty composable, but the seaside components are bound to whatever CSS framework is chosen, there isn't much to do there other than subclassing. This has several drawbacks to integrate code from different teams/projects, because you end up having to use a whole set of classes from the "foreign" library (e.g. QCMagritte) rather than just a few components. One thing that I disliked is the fact that from Magritte 2 to 3 the descriptions moved from the class side to the instance side, to allow dynamic containers based on instance state/behavior, this is really powerful, but I'd preferred a two step lookup to build the instances, it is... look at the instance side first, and then class side, this way you can manipulate description without having to instantiate a sample instance. This is mandatory if you're building an UI like a reporting, querying tool where do not know in advance which instances are going to be returned. However I think this is a discussion for the Magritte, or Seaside, mailing list. Regards, -- Esteban

March 20, 2017

Re: [Pharo-users] OSProcess or OSSubprocess?

by Alistair Grant

Hi Mariano, On 20 March 2017 at 12:24, Mariano Martinez Peck <marianopeck(a)gmail.com> wrote: > > .... > > Hi Hernan, > > There is a tail -f example in OSSubprocess documentation that prints the > result on an inpsector and updates it [1]. However, OSSubprocess has > problems on Linux with the default VM... you must use the threaded heartbeat > (not the itimer).. see recent threads discussions for this. > > Let me know if that works. > > > [1] > https://github.com/marianopeck/OSSubprocess#processing-streams-while-running Cool! Thanks for pointing this out. I've just hit the same requirement, and this will also prompt me to move to the threaded VM. Cheers! Alistair

March 20, 2017

Re: [Pharo-users] looking for tools to manually select ast nodes for metalinks

by Steven Costiou

Hi Ben, Guille, thanks !!! I will look into that =) Steven. Le 2017-03-20 10:39, Guillermo Polito a Ã©crit : > Hi Steven, > > Yes there is! It's called "Suggestions". Suggestions is a project started some years ago by Gisela Decuzzi as an intern of RMoD. The main idea is to propose actions based on the AST corresponding to the cursor. This means that given the position of the cursor, suggestions already: > > - parses the code > - selects the corresponding node > - provides possible actions for the node. E.g., > - refactorings > - code navigation > - breakpoints and watchpoints (through Metalinks) > > Check the SmartSuggestions package. > > Guille > > On Mon, Mar 20, 2017 at 1:12 AM, Ben Coman <btc(a)openinworld.com> wrote: > >> On Mon, Mar 20, 2017 at 4:06 AM, Steven Costiou <steven.costiou(a)kloum.io> wrote: >>> Hi, >>> >>> i would like a way to define metalinks at debug time, by "visually" >>> selecting ast nodes in the debugger. For example i would like to see a >>> method that i am debugging and by a clic on the source code it would select >>> the associated ast node and allow me to put a metalink right there. >>> >>> - does that exist ? >> >> This might get you close... >> >> Debug into... >> 1 printString >> >> In Variables tab, select "thisContext" >> >> In Raw tab, select method >> >> In AST tab, select your node. >> >> From there on, I don't know. Please report back if you get something to work. >> >>> >>> - if not, does a tool to display or to navigate through metalinks already >>> exists ? >> >> I don't think so. Pharo 6 has only the first step of having the basic >> underlying system implemented. >> >> cheers -ben >> >>> >>> - if not any, can somebody point me to documentation and/or ressources to >>> extend the debugger so i can try to build it ? (i've already seen the spec >>> book to build ui but i want to see what is possible with the debugger) -- kloum.io

March 20, 2017

Re: [Pharo-users] OSProcess or OSSubprocess?

by Mariano Martinez Peck

On Mon, Mar 20, 2017 at 6:00 AM, HernÃ¡n Morales Durand < hernan.morales(a)gmail.com> wrote: > Hi guys, > > I need to monitor Unix command execution inside the image. The commands > use heavy computation so they usually report progress in the terminal. > > Are OSProcess or OSSubprocess ready to do in-image monitoring so I can > watch the same terminal output in the Transcript updating for example every > 10 seconds? > > Hi Hernan, There is a tail -f example in OSSubprocess documentation that prints the result on an inpsector and updates it [1]. However, OSSubprocess has problems on Linux with the default VM... you must use the threaded heartbeat (not the itimer).. see recent threads discussions for this. Let me know if that works. [1] https://github.com/marianopeck/OSSubprocess#processing-streams-while-running > Particularly for OSProcess I couldn't find any method to fetch partial > output. > > Cheers, > > HernÃ¡n > > -- Mariano http://marianopeck.wordpress.com

March 20, 2017

Re: [Pharo-users] OSProcess or OSSubprocess?

by Thierry Goubier

2017-03-20 10:00 GMT+01:00 HernÃ¡n Morales Durand <hernan.morales(a)gmail.com>: > Hi guys, > > I need to monitor Unix command execution inside the image. The commands > use heavy computation so they usually report progress in the terminal. > > Are OSProcess or OSSubprocess ready to do in-image monitoring so I can > watch the same terminal output in the Transcript updating for example every > 10 seconds? > > Particularly for OSProcess I couldn't find any method to fetch partial > output. > GitFileTree does fetch partial output over OSProcess, but this is quite a low-level use of OSProcess. What is missing is true pseudo-tty support. Regards, Thierry > > Cheers, > > HernÃ¡n > >

March 20, 2017

Re: [Pharo-users] Magritte extension

by Otto Behrens

Hi, We also made several sub-classes of Magritte classes and / or components. This causes a lot of problems because when we upgraded Magritte it was difficult. It also shows that Magritte (the core classes) cannot change much because the impact will be high for most users of Magritte. This makes us think that Magritte is not easy to use or customise. And it also appears as if all the custom changes are not being generalised into the framework. Perhaps us users of Magritte should stand back and discuss our changes and issues more so that we can improve it? I suppose we need some magritte champion(s) to take it somewhere. To get more flexibility with Magritte styling we are using CSS and classes more in stead of rendering in a different way. Often one can get away with adding a CSS class and changing the style sheet. This is a bit fluffy, but it there's a lot of stuff to talk about here. In principle, are you sharing my frustration with Magritte? Cheers Otto On Sat, Mar 18, 2017 at 7:40 PM, Mariano Martinez Peck <marianopeck(a)gmail.com> wrote: > Hi Asbath, > > Some answers below. But a general comment is that the typical case to > customize magritte is by subclassing and then somehow use your own > subclasses instead of magritte ones. > > On Fri, Mar 17, 2017 at 3:54 AM, Asbath Sama biyalou > <asamabiyalou(a)yahoo.com> wrote: >> >> Hello. >> >> I am using Magritte to generate reports for a pharo class. I want to >> customize generated components in order to have a better design. >> >> My class is Match >> >> Object subclass: #Match >> instanceVariableNames: 'date_match hour_match status actions stadium >> comments teams goals1 goals2 competition' >> classVariableNames: '' >> category: MyProject-Entity' >> >> >> Descriptions for some variables. >> >> Stadium is also a Class. >> >> Match>>descriptionStadium >> <magritteDescription> >> ^ MASingleOptionDescription new >> label: 'Stadium'; >> priority: 305; >> accessor: #stadium; >> options: Team allStadiumsName; >> componentClass: TBSMagritteSelectListComponent; >> beRequired; >> yourself >> >> >> Match>>descriptionTeams >> <magritteDescription> >> ^ MAMultipleOptionDescription new >> label: 'Teams'; >> priority: 805; >> accessor: #teams; >> options: Team allTeamsName; >> componentClass: MAMultiselectListComponent; >> yourself >> >> >> The generated component in the form is not very pretty. I want to have >> multiple checkbox. > > > > In my case, I have customized ALL the components I use for Magritte. That > is, I have my own Fa version (Fa is my package prefix). So I have > FaMAMultiselectListComponent subclass of MAMultiselectListComponent. I have > FaTBSMagritteSelectListComponent, subclass of TBSMagritteSelectListComponent > , etc... so basically, you subclass and override the parent methods you > want. Most common methods you need to override are renderEditorOn: html or > renderViewerOn: html etc...but that depends on WHAT you want to change and > in WHICH component. Note they are all subclasses of MADescriptionComponent. > > You can also set your own properties in the magritte so that you can read > them at your component level. For example, above I could say: > > Match>>descriptionTeams > <magritteDescription> > ^ MAMultipleOptionDescription new > label: 'Teams'; > priority: 805; > accessor: #teams; > options: Team allTeamsName; > propertyAt: 'showInMultipleCheckboxs' put: true; > componentClass: FaMAMultiselectListComponent; > yourself > > > Then in the rendering code of your subclass FaMAMultiselectListComponent you > can do: > > (self magritteDescription propertyAt: 'showInMultipleCheckboxs') ifTrue: [ > self renderAsMultiCheckboxsOn: html... > ] ifFalse: [ > ... > ] > > Of course, you can even define the accessors for #showInMultipleCheckboxs as > extension methods in MADescription > > > To avoid having to define FaMAMultiselectListComponent for each > MAMultipleOptionDescription .. well, there are many ways. But I guess that > is for a later topic. Ask me if you want this. > > >> >> >> Match>>descriptionStatus >> <magritteDescription> >> ^ MASingleOptionDescription new >> label: 'Status'; >> priority: 505; >> accessor: #status; >> options: #('Pas en cours' 'En Cours' 'Mi-temps' 'TerminÃ©'); >> componentClass: TBSMagritteSelectListComponent; >> yourself >> >> >> For the report I use the class SEMatchReport >> >> TBSMagritteReport subclass: #SEMatchReport >> instanceVariableNames: 'report' >> classVariableNames: '' >> category: 'MyProject-Components' >> >> >> SEMatchReport class>>from >> | report matchs commandColumn | >> matchs := Match selectAll. >> report:= self rows: matchs description: (self >> filteredDescriptionsFrom: matchs anyOne). >> >> commandColumn := MACommandColumn new. >> >> (SEMatchReport new session isSimpleAdmin) >> ifTrue: [ >> report addColumn: (commandColumn >> addCommandOn: report selector: #editMatch: text: 'Modifier'; >> yourself; >> addCommandOn: report selector: #addAction: text: 'Ajouter >> une action'; yourself ) ] >> ifFalse: [ >> report addColumn: (commandColumn >> addCommandOn: report selector: #editMatch: text: 'Modifier'; >> yourself; >> addCommandOn: report selector: #deleteMatch: text: >> 'Supprimer'; yourself; >> addCommandOn: report selector: #addAction: text: 'Ajouter >> une action'; yourself ) ]. >> >> ^ report >> >> I want to put some bootstrapp icons near commandColumn actions. like >> pencil or delete on 'Supprimer'. >> > > > To do this, again, subclass MACommandColumn, override renderCellContent:on: > and use your particular subclass. > > >> >> >> SEMatchReport class>>filteredDescriptionsFrom: aMatch >> ^ aMatch magritteDescription select: [ :each | #(date_match >> hour_match stadium goals1 goals2 teams) >> includes: each accessor selector ] >> >> I notice that on the generated report on column stadium we have >> aStadium. We don't have the value (the name). I don't know how to >> mention it in the filtered descriptions. > > > > That's weird. I don't know. > >> >> >> Change the color of lines in the report, etc. All that things. >> > > In this case I don't use the magritte row/even css classes. > > In my case, I have a subclass of MAReport (I guess you can subclass from the > Bootstrap one) and changed the table rendering to use: > > table > beHover; > beBordered; > beCondensed; > beStriped; > > etc... so I let colors to Bootstrap / CSS. > > > Cheers, > > >> >> Sorry for the bad english. >> >> Thanks. >> >> Asbath >> > > > > -- > Mariano > http://marianopeck.wordpress.com

March 20, 2017

Re: [Pharo-users] looking for tools to manually select ast nodes for metalinks

by Guillermo Polito

Hi Steven, Yes there is! It's called "Suggestions". Suggestions is a project started some years ago by Gisela Decuzzi as an intern of RMoD. The main idea is to propose actions based on the AST corresponding to the cursor. This means that given the position of the cursor, suggestions already: - parses the code - selects the corresponding node - provides possible actions for the node. E.g., - refactorings - code navigation - breakpoints and watchpoints (through Metalinks) Check the SmartSuggestions package. Guille On Mon, Mar 20, 2017 at 1:12 AM, Ben Coman <btc(a)openinworld.com> wrote: > On Mon, Mar 20, 2017 at 4:06 AM, Steven Costiou <steven.costiou(a)kloum.io> > wrote: > > Hi, > > > > i would like a way to define metalinks at debug time, by "visually" > > selecting ast nodes in the debugger. For example i would like to see a > > method that i am debugging and by a clic on the source code it would > select > > the associated ast node and allow me to put a metalink right there. > > > > - does that exist ? > > This might get you close... > > Debug into... > 1 printString > > In Variables tab, select "thisContext" > > In Raw tab, select method > > In AST tab, select your node. > > From there on, I don't know. Please report back if you get something to > work. > > > > > > - if not, does a tool to display or to navigate through metalinks already > > exists ? > > I don't think so. Pharo 6 has only the first step of having the basic > underlying system implemented. > > cheers -ben > > > > > - if not any, can somebody point me to documentation and/or ressources to > > extend the debugger so i can try to build it ? (i've already seen the > spec > > book to build ui but i want to see what is possible with the debugger) > >

March 20, 2017