September 2018 - Pharo-dev - lists.pharo.org

Re: [Pharo-dev] Pharo Downloads are sluggish

by Esteban Maringolo

I understand, I'm a heavy user of P2P for most of what's possible, even have a , but HTTP CDNs exists for a reason, which is "distributed availability" and low latency. If we add a bitorrent client as a requirement to have reasonable download speed then you're adding one extra step to the whole onboarding experience. Maybe instead of using a CDNs we can think of using mirrors, but then if the mirror can't withstand the load, you end up in the same situation. Regards, Esteban A. Maringolo El sÃ¡b., 22 sept. 2018 a las 2:25, ponyatov (<dponyatov(a)gmail.com>) escribiÃ³: > I say not about size but *distributed availability*. > > Community members can hold many torrent peers let anybody download > minimizing problems with networks, servers etc, see the topic title. > > PS: maybe in 2048 people will have distributed Smalltalk, but not in this > life with the image hell > > > > -- > Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html > >

Sept. 22, 2018

[Pharo 7.0-dev] Build #1261: Bug fix for bindings to inherited pool dictionaries (issue 22470)

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #1261 was: FAILURE. The Pull Request #1816 was integrated: "Bug fix for bindings to inherited pool dictionaries (issue 22470)" Pull request url: https://github.com/pharo-project/pharo/pull/1816 Issue Url: https://pharo.fogbugz.com/f/cases/22470 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

Sept. 22, 2018

Re: [Pharo-dev] Pharo Downloads are sluggish

by ponyatov

I say not about size but *distributed availability*. Community members can hold many torrent peers let anybody download minimizing problems with networks, servers etc, see the topic title. PS: maybe in 2048 people will have distributed Smalltalk, but not in this life with the image hell -- Sent from: http://forum.world.st/Pharo-Smalltalk-Developers-f1294837.html

Sept. 22, 2018

[Pharo 7.0-dev] Build #1260: 22432 New Release test: NoUnusedTemporaryVariablesLeftTest

by ci-pharo-ci-jenkins2＠inria.fr

There is a new Pharo build available! The status of the build #1260 was: FAILURE. The Pull Request #1782 was integrated: "22432 New Release test: NoUnusedTemporaryVariablesLeftTest" Pull request url: https://github.com/pharo-project/pharo/pull/1782 Issue Url: https://pharo.fogbugz.com/f/cases/22432 Build Url: https://ci.inria.fr/pharo-ci-jenkins2/job/Test%20pending%20pull%20request%2…

Sept. 21, 2018

Re: [Pharo-dev] Mailing list Archive - Bandwidth Limit Exceeded

by Marcus Denker

Fixed. > On 21 Sep 2018, at 09:50, Marcus Denker <marcus.denker(a)inria.fr> wrote: > > I will check with support. > >> On 21 Sep 2018, at 09:49, Torsten Bergmann <astares(a)gmx.de> wrote: >> >> Hi Marcus, >> >> I receive a "Bandwidth Limit Exceeded - The server is temporarily unable to service your request due to the site owner reaching >> his/her bandwidth limit. Please try again later." message on http://lists.pharo.org >> >> If I remember correctly we already had this one or two years ago. >> >> Thanks >> T. >

Sept. 21, 2018

Re: [Pharo-dev] Mailing list Archive - Bandwidth Limit Exceeded

by Marcus Denker

I will check with support. > On 21 Sep 2018, at 09:49, Torsten Bergmann <astares(a)gmx.de> wrote: > > Hi Marcus, > > I receive a "Bandwidth Limit Exceeded - The server is temporarily unable to service your request due to the site owner reaching > his/her bandwidth limit. Please try again later." message on http://lists.pharo.org > > If I remember correctly we already had this one or two years ago. > > Thanks > T.

Sept. 21, 2018

Mailing list Archive - Bandwidth Limit Exceeded

by Torsten Bergmann

Hi Marcus, I receive a "Bandwidth Limit Exceeded - The server is temporarily unable to service your request due to the site owner reaching his/her bandwidth limit. Please try again later." message on http://lists.pharo.org If I remember correctly we already had this one or two years ago. Thanks T.

Sept. 21, 2018

Deprecation guide for methods, classes and packages

by Torsten Bergmann

Hi, >From Bug #20900/ PR 1818 it looked like it is not really clear on how to "deprecate" correctly. Therefore I summarized and wrote a short guide following what we contributors used as best practice so far. Also read [1], [2] and [3] and check some of the deprecation settings in the Pharo settings browser. Just open SettingsBrowser and search for "Depre". I write this mail to "pharo-dev" with CC: "pharo-users" list - because of common interest. But comments and replies should please go to "pharo-dev" so we do not disrupt the thread too much. Thanks! Please review what I wrote and ideally we can put the text or a link to it as "Deprecation guide" into the documentation / wiki / ... on GitHub afterwards. Hope this helps a little bit. Thanks T. [1] https://en.wikipedia.org/wiki/Deprecation [2] https://www.peteruhnak.com/blog/2015/08/06/transforming-deprecated-code-wit… [3] https://pharoweekly.wordpress.com/2016/06/17/automatic-deprecation-refactor… ------------------------------------------------------------------------------------------------------------------------------ In any Pharo version we have a special package hosting deprecated methods and classes. This allows smooth migration for Pharo users as code should still load and runs to some extent - but deprecation warnings and possibly a helping hand (in form of automatic code transformations) could be given to support migration. As we currently work on Pharo7 there is a "Deprecated70" package in the image hosting deprecated methods and classes intended to be removed for Pharo 8. When Pharo 7 is released the users can easily spot the deprecations - also because they are striked through (more on this below). When Pharo 8 iteration starts the "Deprecated70" will be replaced with an empty "Deprecated80" package. DEPRECATING METHODS =================== If you deprecate a METHOD use #deprecated: message and switch it to be an extension method in *Deprecated70*. If the method could be replaced by calling another method then dispatch the call to the new method and ideally provide a transformation so next time the old method is called the sender method is automagically rewritten and could be adopted/migrated: oldMethod "An old method we do not use anymore. But luckily there is a replacement." self deprecated: 'Please use #newMethod instead' transformWith: '`@receiver oldMethod' -> '`@receiver newMethod'. ^ self newMethod So if there is a method like foo self oldMethod sending the old deprecated method it will be migrated to foo self newMethod as soon as foo, the oldMethod and the transformation is called. If the automatically migrated method is under version control (Monticello or git) you will easily see that #foo was modified. In case there is no replacement for a method you would like to deprecate then it is best to throw an error when it is used: veryOldMethod "A very old method hopefully nobody uses. So we do no provide a replacement." self deprecated: 'The method #veryOldMethod is deprecated without any replacement'. ^ self error: 'Do not use #veryOldMethod anymore' Note that the method is striked through in the Pharo system browser in both cases to show that it should not be used anymore. In the standard image calling a deprecated method is logged on the Transcript and flagged with a warning dialog. This should help users and maintainers of external packages to easily spot when they still use deprecated code. You can change these settings in Pharo Settings Browser under Tools -> Debugging -> Deprecation Handling. DEPRECATING CLASSES =================== If you deprecate a CLASS implement #isDeprecated on class side to return true and move it to package "Deprecated70" Ideally write in the class comment why it is deprecated and if there is a replacement class what to use instead Object subclass: #OldClass instanceVariableNames: '' classVariableNames: '' package: 'Deprecated70-OldTag' and comment to explain the reasons: OldClass comment: 'Use NewClass instead'. and implement in class side method category 'testing' isDeprecated "We do not need this class anymore" ^true Note that the class is striked through in the system browser. DEPRECATING PACKAGES ==================== If you deprecate a PACKAGE then rename it with "Deprecated" in the name. Let the class side of the package manifest (typically a subclass of PackageManifest within the package) als return #isDeprecated with true. In the class comment of the package manifest (which also serves as package comment) explain why the package is deprecated and if there is a replacement or not. Example: your package name is "OldPackage-Core", so the new name is "OldPackage-Deprecated-Core". Create a manifest or use the existing one. A Manifest should be a subclass of PackageManifest and should be categorized within a 'Manifest' tag. PackageManifest subclass: #ManifestOldPackageCore instanceVariableNames: '' classVariableNames: '' package: 'OldPackage-Deprecated-Core-Manifest' and comment to explain the reasons: ManifestOldPackageCore comment: 'This package is not useful anymore. Check NewPackage for replacement'. and implement in class side method category 'testing' isDeprecated "We do not need this package anymore" ^true Note that the manifest class and the package is striked through in the system browser. Be aware that a deprecated package might be removed completely already in the next Pharo major version iteration.

Sept. 21, 2018

Re: [Pharo-dev] Debugger should step silently over #halt

by Ben Coman

On Fri, 21 Sep 2018 at 01:51, Tim Mackinnon <tim(a)testit.works> wrote: > I never knew it did that - but it makes sense. Presumably pressing resume > then activated them again right? > yes. Though IIUC its not so much that they were deactivated, but more like the system goes... I hit a halt, hey I'm already halted, so no-op. cheers -ben > > Tim > > Sent from my iPhone > > On 20 Sep 2018, at 08:42, Ben Coman <btc(a)openinworld.com> wrote: > > I'll often put several breakpoints in code like this... > ``` > debug := true. > "Somewhere else in application code..." > debug ifTrue: [ self halt ]. > 1 + 1. "Representing some longer bit of application code" > debug ifTrue: [ self halt ]. > 2 + 2. "Representing some longer bit of application code" > debug ifTrue: [ self halt ]. > 3 + 3. "Representing some longer bit of application code" > ``` > While stepping <Over> code after hitting the first halt, > I am used to subsequent halts being silently stepped over. > I just confirmed this with > http://files.pharo.org/platform/Pharo6.0-win.zip > and also build 60541 started from PharoLauncher. > > However build 71231 and build 60542 do not silently step over a #halt. > Each produces slightly different results (pics attached). > <P01-Double-Halt-Pharo7.png> > <P02-Double-Halt-Pharo60542.png> > > I've logged an issue... > > https://pharo.fogbugz.com/f/cases/22474/Debugger-should-step-silently-over-… > > cheers -ben > >

Sept. 20, 2018

Re: [Pharo-dev] Debugger should step silently over #halt

by Tim Mackinnon

I never knew it did that - but it makes sense. Presumably pressing resume then activated them again right? Tim Sent from my iPhone > On 20 Sep 2018, at 08:42, Ben Coman <btc(a)openinworld.com> wrote: > > I'll often put several breakpoints in code like this... > ``` > debug := true. > "Somewhere else in application code..." > debug ifTrue: [ self halt ]. > 1 + 1. "Representing some longer bit of application code" > debug ifTrue: [ self halt ]. > 2 + 2. "Representing some longer bit of application code" > debug ifTrue: [ self halt ]. > 3 + 3. "Representing some longer bit of application code" > ``` > While stepping <Over> code after hitting the first halt, > I am used to subsequent halts being silently stepped over. > I just confirmed this with http://files.pharo.org/platform/Pharo6.0-win.zip > and also build 60541 started from PharoLauncher. > > However build 71231 and build 60542 do not silently step over a #halt. > Each produces slightly different results (pics attached). > <P01-Double-Halt-Pharo7.png> > <P02-Double-Halt-Pharo60542.png> > > I've logged an issue... > https://pharo.fogbugz.com/f/cases/22474/Debugger-should-step-silently-over-… > > cheers -ben

Sept. 20, 2018