February 2016 - Pharo-dev - lists.pharo.org

Re: [Pharo-dev] MetaclassTest>>testMetaclassAndTraitClassRespectsPolymorphismRules

by Henrik Nergaard

The issue should be unrelated: https://pharo.fogbugz.com/f/cases/17616/CI-Server-Failing-tests-related-to-… Best regards, Henrik -----Original Message----- From: Pharo-dev [mailto:pharo-dev-bounces@lists.pharo.org] On Behalf Of Sven Van Caekenberghe Sent: Saturday, February 20, 2016 11:05 AM To: Pharo Development List <pharo-dev(a)lists.pharo.org> Subject: [Pharo-dev] MetaclassTest>>testMetaclassAndTraitClassRespectsPolymorphismRules Hi, I am trying to load a new version of STON in Pharo 5 https://pharo.fogbugz.com/f/cases/16893/Update-STON One of the tests that fails is MetaclassTest>>testMetaclassAndTraitClassRespectsPolymorphismRules More specifically, (Metaclass localSelectors difference: ClassTrait localSelectors) is no longer equal to the exceptions defined in the test because #stonName and #stonOn: were added to Metaclass What should I do ? Did I do something wrong ? Should I add to the exceptions defined in the test or should I add them to ClassTrait ? I think the first sounds the most correct. Thx, Sven

Feb. 20, 2016

Re: [Pharo-dev] Any download statistics for Pharo

by Sven Van Caekenberghe

There used to be some in the past, but beware: CI jobs download Pharo *a lot* ;-) > On 20 Feb 2016, at 11:15, Damien Cassou <damien.cassou(a)inria.fr> wrote: > > Hi, > > do we have any download statistics for Pharo? > > > Best, > > -- > Damien Cassou > http://damiencassou.seasidehosting.st > > "Success is the ability to go from one failure to another without > losing enthusiasm." --Winston Churchill >

Feb. 20, 2016

Re: [Pharo-dev] GTSpotter, the 95% case

by Sven Van Caekenberghe

> On 20 Feb 2016, at 11:06, stepharo <stepharo(a)free.fr> wrote: > > Tx sven > I will see if I use this scenario for a video to explain my friend Spotter. > If you have scenario that I can blindly follow I'm eager to try to produce a video. Yes, that would be good. But even better would be if one starts to use it a bit in day to day routine, only then do you really feel it, and then we can iterate and try to improve (I must say that I learned new stuff while writing this mail ;-) There should be something like on OS X > System Preferences > Trackpad where you have animations and videos explaining thing in the same UI (not jumping to something else). At least, there should be a cookbook explaining really well, high level, for beginners, how to use Spotter (but things have to be stable). >> Hi, >> >> >> This is in the latest #50591, BTW. >> >> Starting with the 3rd one, searching for classes, that has been working well since day one and has benefited from general improvements. >> >> 'zncl' gives me ZnClient immediately. >> >> Yes, you have to learn how to dive into a result to expand it, but that is not often needed in this case. >> >> 'readstream', then first arrow (or cmd-shit-right), then 'co' gives me ZnCodePointReadStream and ZnEncodeReadStream. >> >> In both cases, just typing enter browses the class, which is what I want. >> >> Both implementers and senders do work well today, you just have to figure out how (learnability is an issue, but not everything can be obvious in this world). >> >> 'get: #i' but also, 'get:' + cmd-m gives you all implementors of #get: and spotter automatically expands from a limited set of 5 matches to 100 > > How do I get to see all of them and not just the 100 first. > When I click on the arrow I get the same list 100/127 but not all Yes, that is true, 100 is probably some arbitrary limit In the past there was just one limit for all categories, maybe it is different now, it could be much large or infinite in some cases The idea is to keep refining the search (like I described below) >> Enable preview (click outer arrow, or cmd-p) and you see the implementors right there >> >> Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnClient>>#get: >> >> Type enter to open the method in a browser >> >> Similar approach for senders, using #s and cmd-n >> >> 'join: #s'but also, 'join:' + cmd-n gives you all senders of #join: and spotter automatically expands >> >> Enable preview (click outer arrow, or cmd-p) and you see the implementors right there >> >> Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnDefaultServerDelegate>>#welcomePageCss >> >> Type enter to open the method in a browser >> >> Class references, I don't know how to do that, but I can't see why it would not fit and it would be a good addition. >> >> Spotter adds to the other ways of doing these 4 actions, the old one still work, as does command-click (one of my favourites). >> >> Sven >> >> PS: I don't wan to narrow down the discussion to just these 4 cases, Spotter and GT are about way more than just this, you can search for many, many things (for example, type 'json' and you can install 3 packages from the catalog), and you can extend these tools yourself. I also know that things must improve further.

Feb. 20, 2016

Any download statistics for Pharo

by Damien Cassou

Hi, do we have any download statistics for Pharo? Best, -- Damien Cassou http://damiencassou.seasidehosting.st "Success is the ability to go from one failure to another without losing enthusiasm." --Winston Churchill

Feb. 20, 2016

TaskIT chapter

by Yuriy Tymchuk

HI, here it says that there is a chapter about it in Pharo Enterprise book: https://pharoweekly.wordpress.com/2014/06/23/taskit-handling-processes-nice… <https://pharoweekly.wordpress.com/2014/06/23/taskit-handling-processes-nice…> But I cannot find it. Can someone give me a pointer? Cheers. Uko

Feb. 20, 2016

Re: [Pharo-dev] GTSpotter, the 95% case

by stepharo

Tx sven I will see if I use this scenario for a video to explain my friend Spotter. If you have scenario that I can blindly follow I'm eager to try to produce a video. > Hi, > > > This is in the latest #50591, BTW. > > Starting with the 3rd one, searching for classes, that has been working well since day one and has benefited from general improvements. > > 'zncl' gives me ZnClient immediately. > > Yes, you have to learn how to dive into a result to expand it, but that is not often needed in this case. > > 'readstream', then first arrow (or cmd-shit-right), then 'co' gives me ZnCodePointReadStream and ZnEncodeReadStream. > > In both cases, just typing enter browses the class, which is what I want. > > Both implementers and senders do work well today, you just have to figure out how (learnability is an issue, but not everything can be obvious in this world). > > 'get: #i' but also, 'get:' + cmd-m gives you all implementors of #get: and spotter automatically expands from a limited set of 5 matches to 100 How do I get to see all of them and not just the 100 first. When I click on the arrow I get the same list 100/127 but not all > > Enable preview (click outer arrow, or cmd-p) and you see the implementors right there > > Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnClient>>#get: > > Type enter to open the method in a browser > > Similar approach for senders, using #s and cmd-n > > 'join: #s'but also, 'join:' + cmd-n gives you all senders of #join: and spotter automatically expands > > Enable preview (click outer arrow, or cmd-p) and you see the implementors right there > > Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnDefaultServerDelegate>>#welcomePageCss > > Type enter to open the method in a browser > > Class references, I don't know how to do that, but I can't see why it would not fit and it would be a good addition. > > Spotter adds to the other ways of doing these 4 actions, the old one still work, as does command-click (one of my favourites). > > Sven > > PS: I don't wan to narrow down the discussion to just these 4 cases, Spotter and GT are about way more than just this, you can search for many, many things (for example, type 'json' and you can install 3 packages from the catalog), and you can extend these tools yourself. I also know that things must improve further. > > > > > >

Feb. 20, 2016

[Pharo-dev] MetaclassTest>>testMetaclassAndTraitClassRespectsPolymorphismRules

by Sven Van Caekenberghe

Hi, I am trying to load a new version of STON in Pharo 5 https://pharo.fogbugz.com/f/cases/16893/Update-STON One of the tests that fails is MetaclassTest>>testMetaclassAndTraitClassRespectsPolymorphismRules More specifically, (Metaclass localSelectors difference: ClassTrait localSelectors) is no longer equal to the exceptions defined in the test because #stonName and #stonOn: were added to Metaclass What should I do ? Did I do something wrong ? Should I add to the exceptions defined in the test or should I add them to ClassTrait ? I think the first sounds the most correct. Thx, Sven

Feb. 20, 2016

GTSpotter, the 95% case

by Sven Van Caekenberghe

Hi, There are recurring questions about GTSpotter, along the lines of: why Spotter does not work well for 95% of the cases - implement - send - class - class refs? This is in the latest #50591, BTW. Starting with the 3rd one, searching for classes, that has been working well since day one and has benefited from general improvements. 'zncl' gives me ZnClient immediately. Yes, you have to learn how to dive into a result to expand it, but that is not often needed in this case. 'readstream', then first arrow (or cmd-shit-right), then 'co' gives me ZnCodePointReadStream and ZnEncodeReadStream. In both cases, just typing enter browses the class, which is what I want. Both implementers and senders do work well today, you just have to figure out how (learnability is an issue, but not everything can be obvious in this world). 'get: #i' but also, 'get:' + cmd-m gives you all implementors of #get: and spotter automatically expands from a limited set of 5 matches to 100 Enable preview (click outer arrow, or cmd-p) and you see the implementors right there Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnClient>>#get: Type enter to open the method in a browser Similar approach for senders, using #s and cmd-n 'join: #s'but also, 'join:' + cmd-n gives you all senders of #join: and spotter automatically expands Enable preview (click outer arrow, or cmd-p) and you see the implementors right there Now, dive in (first arrow or cmd-shit-right), and type 'zn' and I can see ZnDefaultServerDelegate>>#welcomePageCss Type enter to open the method in a browser Class references, I don't know how to do that, but I can't see why it would not fit and it would be a good addition. Spotter adds to the other ways of doing these 4 actions, the old one still work, as does command-click (one of my favourites). Sven PS: I don't wan to narrow down the discussion to just these 4 cases, Spotter and GT are about way more than just this, you can search for many, many things (for example, type 'json' and you can install 3 packages from the catalog), and you can extend these tools yourself. I also know that things must improve further.

Feb. 20, 2016

Re: [Pharo-dev] Problem loading new class comments

by Marcus Denker

> On 19 Feb 2016, at 22:19, Sven Van Caekenberghe <sven(a)stfx.eu> wrote: > >> >> On 19 Feb 2016, at 22:06, Nicolai Hess <nicolaihess(a)gmail.com> wrote: >> >> >> >> 2016-02-19 15:37 GMT+01:00 Sven Van Caekenberghe <sven(a)stfx.eu>: >> I am working on https://pharo.fogbugz.com/f/cases/16893/Update-STON and I noticed that class comments did not update, even if they are changed. >> >> Specifically, the class comments of ConfigurationOfSton and STONWriter did definitively change, I can see the changed comments in the MC code browser, but when I load the code, the comments stay the same, both in Nautilus as using the #comment message. >> >> #50591 >> >> !!!!! >> There are two changes in the behavior how packages are loaded - somewhere, don't know what exactly. >> See my analysis >> 17655 Changes in package loading behavior >> I don't know how many changed class comments got lost since Pharo-50497 ! > > Thanks for looking into this, Nicolai. > > This strikes me as an important bug, not because I reported it, but because we silently lose code. And because we have to test to guard us against such loss of functionality. > Yes, it seems that for Spur, some code in the class builder was reverted. (e.g. all the code related to migrating classes with Slots) So yes, we have to merge the changes back. Marcus

Feb. 20, 2016

Re: [Pharo-dev] [Pharo-users] World lastKeystroke and openInWorld: aPasteUpMorph

by stepharo

Le 16/6/15 00:34, Nicolai Hess a Ã©crit : > > > 2015-06-11 10:58 GMT+02:00 Markus Schlager <m.slg(a)gmx.de > <mailto:m.slg@gmx.de>>: > > issues opened > > > I added a fix for the broken lastkeystroke,but this will only work for > the ActiveWorld/PasteUpMorph, not for any other > PasteUpMorph. > > With the split of PasteUpMorph in PasteUpMorph and WorldMorph, and the > change of keystroke handling (KMDispatcher), there is more to do to > make this > fully work again - but I don't want to reintroduce the old keyevent > handling just to make this work again. > > I would like to remove all the keystroke handling from > PasteUpMorph/WorldMorph > and use only KMDispatcher. Yessssss > > Maybe there is another(new) way to make this "lastKeystroke"-demo? > You can open a transcript and enable the KMLog > (KMLog setDebug. Transcript open) Yes I do not see why the event handling would have to do something special. > It will log all(!) keyevents and print the matching keymapping actions. > Or something like this: > |m|m :=Morphnew.m openInWorld.m takeKeyboardFocus.m on:($r ctrl)do:[m > color:Colorred].m on:($g ctrl)do:[m color:Colorgreen].m on:($b > ctrl)do:[m color:Colorblue]. > > This will open a morph that changes its color on ctrl+r/g/b > So, maybe we can remove the whole "lastKeystroke" handling and make > demos/examples > like the above one? Yes :) > > > > > Markus > >

Feb. 20, 2016