August 2012 - Pharo-dev - lists.pharo.org

[Pharo-project] New Text Completion suggestions
by Henrik Sperre Johansen Aug. 24, 2012

Aug. 24, 2012

Is it just me who finds the new text completion resulting from O/E merge insufferably disruptive? Whenever I've been editing in an existing method, and try moving back/forward using arrow keys, instead of moving the cursor, it opens a browser on the completion suggestion for whatever I'd just finished writing (if I go right), or does jack squat (if I try moving left)... Ironically, clicking the text editor anywhere BUT the completion dialogue leaves the window open, and starts suggesting for whatever you type at the cursors new location, while clicking ON any part of the popup-pane CLOSES it. moving cursor with the mouse (not that you could before), only thing that works is pressing escape, which is a natural key to reach for whenever you've finished typing a word, no? Not to mention, if there is only one suggestion, and that suggestion you've already typed in completely, it DOES NOT CLOSE. Suggestions: - Don't hijack the arrow keys (at least not left/right) for autocomplete-popup functionality. - Close the fracking dialogue if the only suggestion it has is _exactly what is already typed_ - Handle mouse-clicks on a suggestion by selecting it, clicks anywhere else in the texteditor by closing it. Cheers, Henry PS: AlphaBlendingCanvas suggested for 'aB'... really?

6 13

[Pharo-project] [In Progress] WebSocket support being added to Zinc HTTP Components
by Sven Van Caekenberghe Aug. 24, 2012

Aug. 24, 2012

Yes, the rumors are true: WebSocket support is coming to Zinc HTTP Components and thus to Pharo Smalltalk ! Several people asked for it, but Andy Burnett of Knowinnovation USA provided a financial incentive (thanks a lot Andy for sponsoring open-source Smalltalk software like that), so I am now putting in the effort to make this happen. The goal is to add WebSocket (RFC 6455) client and server support over plain and TLS/SSL streams (ws:// and wss://) to Zinc HTTP Components maintaining compatibility with Pharo 1.3, 1.4 and 2.0, as well as community suported portability to Gemstone/Smalltalk and Squeak. The test for success will be: Use Pharo Smalltalk as both client and server for the Echo example at http://www.websocket.org/echo.html as well as implement the same example directly (through a locally served page). The final code will be open source and MIT licensed. It will include unit tests, documentation as well as basic support on the mailing list. The purpose of this email is twofold: to announce the fact that this is happening and to look for early alpha testers that can help shape the API. Although as yet unfinished, largely undocumented and untested, still containing debugging and tracing code and subject to change, there is already working code that meets the functional goals available in the main development Zn repository. Gofer it url: 'http://mc.stfx.eu/ZincHTTPComponents'; package: 'Zinc-HTTP'; package: 'Zinc-FileSystem'; package: 'Zinc-Tests'; package: 'Zinc-WebSocket-Core'; package: 'Zinc-WebScoket-Tests'; load The first working example is accessing an existing WebSocket test application at http://www.websocket.org/echo.html ZnWebSocketTests>>#testEchoWebSocketsDotOrg | webSocket message | webSocket := ZnWebSocket to: 'ws://echo.websocket.org'. message := 'Greetings from Pharo Smalltalk @ ', TimeStamp now printString. webSocket sendMessage: message. self assert: webSocket readMessage equals: message. webSocket close. The second working example is Zinc implementing the same echo service itself as well as accessing it ZnWebSocketTests>>#testEchoLocally | webSocket message | ZnServer startDefaultOn: 1701. ZnServer default logToTranscript. ZnServer default delegate map: 'ws-test' to: [ :request | ZnResponse ok: (ZnEntity html: ZnWebSocketDelegate wsTestHtml) ]; map: 'ws-test-local' to: [ :request | ZnResponse ok: (ZnEntity html: ZnWebSocketDelegate wsTestLocalHtml) ]; map: 'ws-echo' to: (ZnWebSocketDelegate new prefixFromString: 'ws-echo'; handler: ZnWebSocketEchoHandler new; yourself). webSocket := ZnWebSocket to: 'ws://localhost:1701/ws-echo'. message := 'Greetings from Pharo Smalltalk @ ', TimeStamp now printString. webSocket sendMessage: message. self assert: webSocket readMessage equals: message. webSocket close. ZnServer stopDefault. Where ZnWebSocketEchoHandler>>#value: webSocket "I implement an echo service conversation as a server: reading messages and echoing them back until ConnectionClosed" [ webSocket runWith: [ :message | self crLog: 'Received message: ', message printString, ' echoing...'. webSocket sendMessage: message ] ] on: ConnectionClosed do: [ self crLog: 'Ignoring connection close, done' ] If you select and execute the following part from the above ZnServer startDefaultOn: 1701. ZnServer default logToTranscript. ZnServer default delegate map: 'ws-test' to: [ :request | ZnResponse ok: (ZnEntity html: ZnWebSocketDelegate wsTestHtml) ]; map: 'ws-test-local' to: [ :request | ZnResponse ok: (ZnEntity html: ZnWebSocketDelegate wsTestLocalHtml) ]; map: 'ws-echo' to: (ZnWebSocketDelegate new prefixFromString: 'ws-echo'; handler: ZnWebSocketEchoHandler new; yourself). You can use your browser to access the echo service through the standard JavaScript API http://localhost:1701/ws-test http://localhost:1701/ws-test-local The first URL goes out to websocket.org, the second uses the local Zinc implementation. Again, this is early code, anything might change. So, if you have some prior knowledge in this area and are interested, have a look and let me know what you think. Thx, Sven -- Sven Van Caekenberghe http://stfx.eu Smalltalk is the Red Pill

4 3

Re: [Pharo-project] New Text Completion suggestions
by Fernando Olivero Aug. 23, 2012

Aug. 23, 2012

Igor, as usual your true object-oriented thinking helps us see a better path... We need a customized UI interaction for Smalltalk, instead of asking for TEXT editors devised for a different setup. What's missing is a Smalltalk method shape(morph, etc...), highly customized for editing/navigating statements written in the Smalltalk syntax. ( unless you want to code the methods using the workspace.....) We can still preserve cmd-c for copy, and cmd-v for paste...IMO it's not about inventing a new set of keyboard shortcuts, but more about manipulating objects at the right level of abstraction. Saludos, Fernando On Fri, Aug 24, 2012 at 1:40 AM, Igor Stasenko <siguctua(a)gmail.com> wrote: > On 24 August 2012 01:19, Eliot Miranda <eliot.miranda(a)gmail.com> wrote: >> >> >> On Thu, Aug 23, 2012 at 3:57 PM, Igor Stasenko <siguctua(a)gmail.com> wrote: >>> >>> (a bit orthogonal) >>> i don't understand why we cannot have own, consistent set which is good >>> for us? >>> vim, emacs.. >> >> >> better use sets which are already extremely familiar than invent yet another >> set. to those of us who use these editors (and we are legion) these sets >> have long become almost subconscious to use. >> > > i know that.. i also came to squeak from outside.. > but as we say in Ukraine: don't enter others sanctuary with own > code(set of rules). > >>> why this is so important ? Those editors were not written for editing >>> smalltalk code in mind.. >>> they are best suited for big, hundreds lines of code, files.. >> >> >> they're the two most popular editors of their type. lots of people use them >> for other languages without IDE support. They provide convenient power >> features such as pattern replacement. Not often I find myself filing out >> Smalltalk code and editing it with vim (sadly I've never learned emacs). >> > me too.. i never learned emacs. So, maybe i am completely ignorant and can't see > why it is so important to have it there.. But then, i don't > understand, why most of editors i know > never had emacs/vim key bindings as option? Are they completely stupid? Huh? > > So, you know, if we follow that logic.. hey we don't have a unix command line.. > so maybe we should add an option: either workspace or command line? > and then introduce nice terminal emulation with prompt, users and .bashrc .. > ahh.. what the hell.. lets replicate whole unix environment.. imagine > how many happy users will join us then! > > -- > Best regards, > Igor Stasenko. >

1 0

[Pharo-project] Friday in Ghent
by DeNigris Sean Aug. 23, 2012

Aug. 23, 2012

I'm arriving on Friday morning (about 9am)... anyone else around? Sean

1 0

Re: [Pharo-project] New Text Completion suggestions
by Stephan Eggermont Aug. 23, 2012

Aug. 23, 2012

Doru wrote >I think an option would only add to the confusion. Please replace the >option with one decision. If some do not like it, they can shout and >the designer can take the input into account. But, it's the designer >that decides. +1 Stephan

1 0

[Pharo-project] this is a test
by StÃ©phane Ducasse Aug. 23, 2012

Aug. 23, 2012

1 0

Re: [Pharo-project] [squeak-dev] ObjectAsMethodWrapper >> #uninstall breaks class organisation
by Frank Shearar Aug. 23, 2012

Aug. 23, 2012

On 23 August 2012 12:49, Bert Freudenberg <bert(a)freudenbergs.de> wrote: > On 2012-08-23, at 11:24, Frank Shearar wrote: > >> It looks like PackageInfo possibly isn't catching the AddEvent sent by >> ClassDescription >> #addSelector:withMethod:notifying: but I'm just >> guessing: I don't know enough about this part of the system. > > > No, PackageInfo is not event-based. It does not "store" classes or methods. Instead, when you ask it for classes and methods in the package, it looks at the system. > > What must be happening is that the class's organizer gets out-of-sync with the method dictionary. The System Browser only looks at the organization to display methods. So if a selector is not in the organizer, the Browser does not show it at all. OK. So the organizer loses the reference because ObjectAsMethodWrapper calls #removeSelectorSilently:, which eventually calls ClassDescription >> #removeSelector:. It re-adds the now-unwrapped CompiledMethod to the method dictionary with #addSelector:withMethod: which eventually calls ClassDescription >> #basicAddSelector:withMethod:. However, this does _not_ add the selector to the organization's elementArray, and Bad Things Happen. I'm trying to find how things get _into_ a Categorizer's elementArray. I suspect it's via ClassDescription >> #addAndClassifySelector:withMethod:inProtocol:notifying:. A Browser would do this via accepting code, while if you're hacking on the method dictionary directly, you'll need to do this yourself. This looks like it would do the trick. At least, it makes the failing test pass: ObjectAsMethodWrapper >> uninstall | category | oldMethod isNil ifTrue: [^self]. category := wrappedClass organization categoryOfElement: selector. "<-- store the original category" wrappedClass removeSelectorSilently: selector. wrappedClass addSelector: selector withMethod: oldMethod. wrappedClass organization classify: selector under: category. "<-- categorise the restored method." frank > This is different in Monticello - if you click "browse" in MC it makes a package snapshot by asking PackageInfo, which iterates through the method dictionary. After running the test this blows up, because it finds the "bar" method, but that is not in the organizer, so its category is nil and you get a walkback. > > Events are only used to mark the Monticello working copy dirty. Not to actually keep track of method additions. > > - Bert - > > >

1 0

[Pharo-project] [update 2.0] #20257
by Esteban Lorenzano Aug. 23, 2012

Aug. 23, 2012

20257 ----- Issue 3459: includesSubstring: aString caseSensitive: caseSensitive / includesSubString: http://code.google.com/p/pharo/issues/detail?id=3459 Issue 3518: #warningAllowed does not handle UndeclaredVariableWarning http://code.google.com/p/pharo/issues/detail?id=3518 Issue 6435: Spotlight gives an error message when asked for World: PasteUpMorph asNode DNU. http://code.google.com/p/pharo/issues/detail?id=6435 Issue 6501: Morph>>drawOnCanvas: does not seems used http://code.google.com/p/pharo/issues/detail?id=6501 Issue 6565: Fix #next:putAll:startingAt: for MultiByteFileStream http://code.google.com/p/pharo/issues/detail?id=6565

2 1

[Pharo-project] Pharo 2.0 #20267
by Sven Van Caekenberghe Aug. 23, 2012

Aug. 23, 2012

Hi Guys, I like to report postive things from time to time, mailing lists shouldn't be about bugs and problems only. I just took a fresh 2.0 image (#20267) and just for fun tried running all tests on my machine, something that didn't work very well in the past. I am happy to report that the _all_ passed ! Great work. All these small and big steps do make a difference. Sven

1 0

[Pharo-project] New Text Completion suggestions
by Torsten Bergmann Aug. 23, 2012

Aug. 23, 2012

1. In Pharo if you expand a keyword message like #replaceAll:with: the cursor is set already to the first argument (so one can continue typing) but to enter the second argument one has to move left using arrow keys "manually". How often you have to move left is dependent on the size of the second word part of the message selector. Why not just TAB to set the cursor to the next argument. 2. What I also really miss compared to other IDE's for completion are "custom completion templates". So in Eclipse by default there is a "sysout" template. It's enough to write "sys" and by expanding this ends up in System.out.println("") where the cursor is already between the double quotes to continue typing. The templates can be defined using placeholders and TAB brings you to the next one to be filled out. This video demonstrated this (at around 4:20): http://www.youtube.com/watch?v=kqXUMhRkn9g 3. AFAIK VisualWorks is modified to also have "non-Smalltalk" code on methods - so you can store resources on methods and the editor can be different depending on content type. I would like to see Pharo having a similar feature to better support non-Smalltalk content within the image. So one can imagine a styled CSS editor for Seaside resources with CSS tag completion, XML editor with completion depending on XSL type definitions, SQL editor with completion, ... Not that we should support all these completion types/grammars but at least have an open framework to plugin such mechanisms in custom editors/completions. I know this is a lot of work ... but as a suggestion we should discuss if this is a direction we want to go (or just have a good Smalltalk editor/completion). Bye T.

2 1