June 2020 - Pharo-dev - lists.pharo.org

Re: [Pharo-dev] OpenSSL binding Re: About strange email related to smalltalkhub read-only on squeak-dev

by Pierce Ng

On Thu, Jun 04, 2020 at 10:31:09AM +0200, Norbert Hartl wrote: > > Am 04.06.2020 um 02:44 schrieb Pierce Ng <pierce(a)samadhiweb.com>: > > On Tue, Jun 02, 2020 at 11:17:30AM +0200, Norbert Hartl wrote: > >> PierceNg has an implementation that implements a subset of openssl. > >> This implementation is modeled after the library so lots of class > >> methods. I'd prefer to have something more object model like. > > > > A very small subset currently, as my original need was to create an X509 > > request in code. PRs welcome. > > > > https://github.com/PierceNg/OpenSSL-Pharo > > > If it is a small subset it might be feasible to talk about the approach taken. Sure. I don't see too many misplaced class methods myself. I've just loaded the package into a fresh image for a spin. My self-built VM uses OpenSSL 1.1.1 (instead of OpenSSL 1.0.x in the prebuilt VMs) and there are C API changes between those two OpenSSL versions that break many tests, basic things like XXX_create becoming XXX_new, XXX_init becoming XXX_reset etc. So I've created branches openssl_1_1 and openssl_1_0 that will match the versions used by Pharo. Pierce

June 6, 2020

Re: [Pharo-dev] Discoverability

by Sean P. DeNigris

Sean P. DeNigris wrote > Fixed https://github.com/pharo-project/pharo/pull/6493 ----- Cheers, Sean -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

June 5, 2020

Re: [Pharo-dev] Discoverability

by Sean P. DeNigris

Sean P. DeNigris wrote > the icons and arrows themselves are not properly indented (they are all in > one vertical > line regardless of depth in tree) :/ Fixed: `displayMainColumnBy: [ :cell :item | cell useFullIndentation... ]` ----- Cheers, Sean -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

June 5, 2020

Re: [Pharo-dev] Discoverability

by Sean P. DeNigris

Denis Kudriashov wrote > Try to replace... with... Thanks!! That added tree node arrows to the class tree :) but the icons and arrows themselves are not properly indented (they are all in one vertical line regardless of depth in tree) :/ BTW am I alone in thinking that the class list should be treated as a tree by default? It basically is a tree *already*, but without the ability to collapse/expand which limits understanding a large tree where siblings may be far apart when it's fully expanded. Also, I can't imagine ever coming up with all that on my own, so I'm glad I asked, but I'm worried about the future extensibility/maintenance if original authors aren't available :/ Denis Kudriashov wrote > You were in the right direction. But there is a little detail here. > By default your code would work for domain objects in the list (not a > class > view). > But the problem with classes and methods lists is to support the Ring > model > where classes are not instances of Class and methods are not instances of > CompiledMethod. > My solution was to have a data type for query items. By default it is an > actual class of objects. But for classes and methods I introduced > ClyClass and ClyMethod to address the Ring case. > > So try following in your experiment > ClyClass -> #prepareProjectItemsQueryFrom:in: > And you should have a debugger. > > I wonder, did you try a halt in #treeStructure users . I think it would > help to discover this logic. I checked for senders of #treeStructure (there were none) and then references to the underlying inst var, but for some reason my eyes glossed over `queryToExpand:ifAbsent:` and I only ran across it much later in a debugger, but I don't remember from what route. Although, given your solution above, is this even necessary? ----- Cheers, Sean -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

June 5, 2020

[Pharo 9.0] Build #340: pushOuterVectors-use-tempVectorVarNames

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #340 was: SUCCESS. The Pull Request #6478 was integrated: "pushOuterVectors-use-tempVectorVarNames" Pull request url: https://github.com/pharo-project/pharo/pull/6478 Issue Url: https://github.com/pharo-project/pharo/issues/pushOuterVectors Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

June 5, 2020

[Pharo 8.0] Build #1140: 6418-Backport-5643-into-Pharo8

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #1140 was: SUCCESS. The Pull Request #6429 was integrated: "6418-Backport-5643-into-Pharo8" Pull request url: https://github.com/pharo-project/pharo/pull/6429 Issue Url: https://github.com/pharo-project/pharo/issues/6418 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

June 5, 2020

[Pharo 9.0] Build #339: Created a simple and a complex test for the changes file flushing issue

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #339 was: FAILURE. The Pull Request #6492 was integrated: "Created a simple and a complex test for the changes file flushing issue" Pull request url: https://github.com/pharo-project/pharo/pull/6492 Issue Url: https://github.com/pharo-project/pharo/issues/6491 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

June 5, 2020

[Pharo 9.0] Build #338: 6049-It-should-not-be-possible-to-use-super-as-something-else-than-the-receiver-of-a-message

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #338 was: SUCCESS. The Pull Request #6482 was integrated: "6049-It-should-not-be-possible-to-use-super-as-something-else-than-the-receiver-of-a-message" Pull request url: https://github.com/pharo-project/pharo/pull/6482 Issue Url: https://github.com/pharo-project/pharo/issues/6049 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

June 5, 2020

Re: [Pharo-dev] Discoverability

by Denis Kudriashov

Hi Sean. Ð¿Ñ, 5 Ð¸ÑÐ½. 2020 Ð³. Ð² 16:15, Sean P. DeNigris <sean(a)clipperadams.com>: > I had a "simple" goal: to change Calypso's class pane to a tree. After an > hour or two, I didn't feel any closer! > I implemented it in very early stages. But I did have time to implement it in a better way. The old code is still there. But I did not expose it to the browser widgets API. Try to replace following line in ClyFullBrowser>>switchToFlatClasses classView showQueries: classQueries as: ClyExtensionLastSortedClasses hierarchical with: fullQuery := ClyQuery unionFrom: classQueries as: ClyExtensionLastSortedClasses hierarchical. classView initiateUIChangeBy: [ dataSource := ClyExpandedDataSource on: fullQuery. classView dataSource: dataSource. classView ensureSelectedItemIfNeeded]. The trick is to use ClyExpandedDataSource instead of ClyCollapsedDataSource. The former is responsible to provide expansion to predefined internal tree structure (which you see in default browser). I wanted to merge them but I had no time for that. I thought to myself, > packages are already in a tree, let me dig there! I got a bit worried when > I > saw what seemed to be a mini-DSL for tree making instead of simple message > sends: > ClyFullBrowser >> #newPackageView > ... > treeStructure: { > ClyProjectChildItem -> > #prepareProjectItemsQueryFrom:in:. > RPackage -> #prepareClassGroupQueryFrom:in: }; > But I pressed on and tried to duplicate that logic: > ClyFullBrowser >> #newClassView > ... > ^ self newNavigationView > treeStructure: { > Class -> #prepareProjectItemsQueryFrom:in: } > You were in the right direction. But there is a little detail here. By default your code would work for domain objects in the list (not a class view). But the problem with classes and methods lists is to support the Ring model where classes are not instances of Class and methods are not instances of CompiledMethod. My solution was to have a data type for query items. By default it is an actual class of objects. But for classes and methods I introduced ClyClass and ClyMethod to address the Ring case. So try following in your experiment ClyClass -> #prepareProjectItemsQueryFrom:in: And you should have a debugger. I wonder, did you try a halt in #treeStructure users . I think it would help to discover this logic. Denis ... > > And... nothing. When #prepareClassItemsQueryFrom:in: didn't exist, I got no > debugger. When I created it, placing a halt in it never got triggered. > > Besides my immediate experiment of making the class pane a tree, my more > fundamental questions are: > 1. What is the best heuristic to explore the system? This would be great to > put somewhere publicly. Especially, but as we can see from this post not > exclusively, for new users. > 2. Do we value discoverability? > 3. If so, are we designing our system with this value in mind? > > > > ----- > Cheers, > Sean > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html > >

June 5, 2020

Discoverability

by Sean P. DeNigris

I had a "simple" goal: to change Calypso's class pane to a tree. After an hour or two, I didn't feel any closer! I'm wondering if it's my process or something else. A few questions: 1. Is there a heuristic for finding relevant documentation? We have lots of docs, but there seem to be a lot of separate locations for it. I think I searched on the Nabble ML mirror for "Calypso" posts on either Pharo list. After digging around, I found a discussion where Steph was working on a [] booklet, which I read and was helpful for understanding the model but didn't directly address my issue (not that it should, this is pretty low-level hacking!). Then I tackled the class comments, which are VERY thorough, but it's hard (at least for me) to piece together the overall design from the individual parts. 2. How do we make sophisticated designs discoverable? I remember one of my Aha! moments when I discovered Smalltalk was using Morphic halos on a menu item and easily finding the code it used underneath. Things like that do not generally seem possible anymore in my experience. I thought to myself, packages are already in a tree, let me dig there! I got a bit worried when I saw what seemed to be a mini-DSL for tree making instead of simple message sends: ClyFullBrowser >> #newPackageView ... treeStructure: { ClyProjectChildItem -> #prepareProjectItemsQueryFrom:in:. RPackage -> #prepareClassGroupQueryFrom:in: }; But I pressed on and tried to duplicate that logic: ClyFullBrowser >> #newClassView ... ^ self newNavigationView treeStructure: { Class -> #prepareProjectItemsQueryFrom:in: } ... And... nothing. When #prepareClassItemsQueryFrom:in: didn't exist, I got no debugger. When I created it, placing a halt in it never got triggered. Besides my immediate experiment of making the class pane a tree, my more fundamental questions are: 1. What is the best heuristic to explore the system? This would be great to put somewhere publicly. Especially, but as we can see from this post not exclusively, for new users. 2. Do we value discoverability? 3. If so, are we designing our system with this value in mind? ----- Cheers, Sean -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

June 5, 2020