- Pharo-dev - lists.pharo.org

Esteban's ChangeLog week of 13 March 2017
by estebanlm＠gmail.com March 20, 2017

March 20, 2017

Hello! This is my weekly ChangeLog, from 13 March 2017 to 19 March 2017. You can see it in a better format by going here: http://log.smallworks.eu/web/search?from=13/3/2017&to=19/3/2017 ChangeLog ========= 17 March 2017: -------------- * So, I changed the build process for linux to do what I suggested [here](http://forum.world.st/changing-pharo-packaging-in-linux-td4938906.html). I also prepared the [zeroconf](http://get.pharo.org) scripts, but now I cannot build a VM because there is a problem with latest sources. I hope it will be fixed soon :) * I reverted LD_LIBRARY_PATH I added to [zeroconf](http://get.pharo.org) scripts. There are two reasons for this: * It was bugged. * It will not be necesary as soon as I can build new VMs because we will be distributing a script to execute in linux, with everything prepared there. 15 March 2017: -------------- * I made a new change to [zeroconf](http://get.pharo.org) scripts, this time going back to my location of resources (libraries) on linux. This time, I changed the execution script for linux into something like this: ---- LD_LIBRARY_PATH=$DIR/pharo-vm:$LD_LIBRARY_PATH $DIR/pharo-vm/pharo --nodisplay "$@" ---- this needs to be like that due to the problem with libgit2 dependences, since there is still a possibility that linux will found the incorrect library versions. This matches with the old distribution script, who was adding: ---- LD_LIBRARY_PATH="$PLUGINS:$SVMLLP:${LD_LIBRARY_PATH}" exec $GDB "$BIN/pharo" "$@" ---- so now we will be on track... but still, the "regular" (no zerconf) distribution will not work out of the box, I think we need to change it to the old script in some way. I will think how can we do that without break our current process. 13 March 2017: -------------- * I continued working on [zeroconf](http://get.pharo.org): I added linux threaded VMs. Of course, not as default since it requieres a special configuration, but at least people wanting to use them now can do it. * Continuing with this "administrative" day, I updated [zeroconf](http://get.pharo.org) to use +PharoConsole.exe+ in Windows VM. This will allow windows users to start an image from command line and be able to write to console. Now, to take advantage of this it was necessary to rollback a change I made... like 5 years ago! (it was one of the first workarounds I applied to the image, when we were working on the CI for Pharo). Anyway, the [case: 19832](https://pharo.fogbugz.com/f/cases/19832) is there to fix that. * last post I forget to clarify that I needed to do this update because older VMs didn't work on macOS sierra. * I made some administrative stuff and I updated downloads of Pharo 3.0 and Pharo 4.0 for macOS at [http://files.pharo.org/platform/](http://files.pharo.org/platform/), I also updated the "platform files" for this versions. ... and I also updated the [get-files](http://files.pharo.org/get-files) files for Pharo 3.0. * I started the week tooking some minutes to fix a problem in Catalog I caused when trying to enhance the behaviour between sessions (fix is [case: 19831](https://pharo.fogbugz.com/f/cases/19831)) cheers! Esteban

1 0

64bits Pharo VM for windows
by Nicolas Cellier March 19, 2017

March 19, 2017

Hi, I've built a 64bits pharo.stack.spur VM for windows on my machine, and I'm uploading the changes to opensmalltalk-vm in branch build_pharo_win32_with_cygwin If the appveyor job correctly succeed, I will emit a pull request. The VM does not have the SqueakSSL plugin yet. The 64bits squeak/pharo.cog.spur JIT for windows is still to come, but I did not work on it for a few months... One thing at a time. Let's cross finger Nicolas

2 2

platforms/Nopsys submodule
by Ben Coman March 19, 2017

March 19, 2017

In the network graph hovering over charig/2-March commit I see a new platform Nopsys added for SqueakNOS. [A] https://github.com/OpenSmalltalk/opensmalltalk-vm/network Thats cool !!! It'll be great to see this work with OpenSmalltalk. However I notice its been added as a submodule. [B] https://github.com/charig/opensmalltalk-vm/commit/2bd87b525b6b46439c90506c0… Submodules bring additional complexity, so I thought there should be some discussion on this topic. Here are a couple of balanced introductions to submodules... [C] http://www.catalyst-au.net/blog/git-submodule-misunderstood-beast-or-remors… [D] https://www.philosophicalhacker.com/post/using-git-submodules-effectively/ And a couple of cautions about using them... [E] http://slopjong.de/2013/06/04/git-why-submodules-are-evil/ [F] https://rufflewind.com/2017-02-21/quirks-with-submodules Guido & Javier, Have you used submodules successfully elsewhere? Could you discuss your strategy using a submodule for the platform code rather than adding directly into the opensmalltalk-vm repository? I see your parent repository includes opensmalltalk-vm as a submodule similar to how I *thought* Pharo was doing it, but wait on comment from Esteban. . [G] https://github.com/charig/Nopsys [H] https://github.com/charig/Nopsys/blob/master/.gitmodules So it seems platforms/Nopsys is a sub-sub-module. Esteban, I thought that you were using submodules to incorporate opensmalltlak-vm into the pharo-vm, but I can't see a .gitmodules file there. [I] https://github.com/pharo-project/pharo-vm So did you do it another way, or am I missing something? Could you report your experience with adding opensmalltalk-vm into the pharo-vm repository? cheers -ben

6 8

about propertyAt: and others
by stepharong March 18, 2017

March 18, 2017

Hi I'm looking at cocoon and other property systems. I would like to get a nice library once for all. I want to also revisit Artefact stylesheets. I'm checking the semantics regarding the lookup of a property in a parent when it is not found in the receiver. I have the impression that when there is a parent access and modification should modify the same object: i.e., the parent if the key is found in the parent or the child if the property is found in the child. in cocoon we have a mix of the two semantics and I would like to get your feel about it. propertyAt: aKey "Answer the value of the property ==aKey==, raises an error if the property doesn't exist." ^ self propertyAt: aKey ifAbsent: [ self hasParent ifTrue: [ self parent propertyAt: aKey ] ifFalse: [self errorPropertyNotFound: aKey ]] propertyAt: aKey default: aBlock "Answer the value of ==aKey==. If the key have no entry I look on the parent or execute a block if i'm an orphan." ^ self propertyAt: aKey ifAbsent: [ self hasParent ifTrue: [ self parent propertyAt: aKey default: aBlock ] ifFalse: aBlock ] but propertyAt: aKey ifAbsent: aBlock "Answer the value of the property ==aKey==, or the result of ==aBlock== if the property doesn't exist." ^ self properties at: aKey ifAbsent: aBlock propertyAt: aKey ifAbsentPut: aBlock "Answer the value of the property ==aKey==, or if the property doesn't exist adds and answers the result of evaluating ==aBlock==." ^ self properties at: aKey ifAbsentPut: aBlock In PetitParser there is no lookup. But we can get back to this situation by just not setting a parent. My gut feeling is telling me that overriding on write is not really good because a client of the parent will not see the changes coming from a child. I thaought that stylesheet was supporting lookup but it does not testSubstyleAttributesOverride styleSheet at: #textColor put: #red. styleSheet > #sub > #subsub at: #textColor put: #blue. self assert:( styleSheet get: #textColor ) equals: #red. self assert:( styleSheet > #sub get: #textColor ) equals: #red. self assert:( styleSheet > #sub > #subsub get: #textColor ) equals: #blue. Does not work What you think? Stef

3 2

Script to download Pharo 4 vm for Mac os sierra
by Cyrille Delaunay March 18, 2017

March 18, 2017

Hello, We have a jenkins job, running on a linux machine, that is responsible for building and outputing a Pharo 4 development image + vm ready to be used on mac machines. We use this command line to download the VM and archived it as output of the job: curl -O http://files.pharo.org/vm/pharo/mac/Pharo-VM-mac-latest.zip On my mac (10.12.1), when i open the pharo image with this vm , I am facing what seem to be some known issues for Mac OS sierra: Pharo[13048:431565] *** -[NSPathStore2 stringByAppendingPathExtension:]: cannot append extension 'bundle/Contents/MacOS/' to path ... Now, when I try to download a Pharo 4 vm directly from my mac sierra machine, with a zeroconf script (get.pharo.org/vm40) It works: I am able to open the pharo image with this vm. Is there a script that allows to download a pharo 4 vm working for mac os sierra, from a linux machine ? -- Cyrille Delaunay

5 5

Pharo7 Inspector feature idea - cherry-pick collection items
by Ben Coman March 17, 2017

March 17, 2017

Sometimes when exploring a large data set in inspector, I'd like to cherry pick a few items to act on as a collection. This would need to be valdiated/tested, but It could be useful to be able to multi-select items and have the next pane to the right be the collection containing the list of picked items. For the moment I can get by with... "self first: 5". cheers -ben .

5 6

Fogbugz feedback - "Image Version"
by Ben Coman March 17, 2017

March 17, 2017

A niggle I've had for a while is that the "Image Version" field seems to me to be useless. It defaults to a particular major version. (that even though only a small thing, is one more thing to maintain) It tends to be superseded by the Milestone field. The limited selection of major version is too broad to add any useful info beyond Milestone to narrow down action. So I wonder Image Version if it might be replaced by a freeform field "Image Version" to enter the value from System > About. ----- As an extension, given the flux of the VMs with the move to 64 bit, and ongoing support for both, it might also be good to have a "VM Version" field. Now actually I often find it awkward to report VM version. The text under System > System Reporter > VM General is quite bulky and its not clear what is the really useful parts for identification. So I usually end up pasting the whole thing, but this is too much for an issue tracker field. Also, there is no indication here of itimer versus threaded heartbeats which is sometimes pertinent. Could suitable minimal VM version info be added to System > About so the info required for the issue tracker is all in one simple place? ----- Finally, do we want these version fields to: * remain static for the version initially reported on, or * be refreshed as confirmed in a later version, or * have separate fields "Confirmed in {Image/VM} Version" to track tests. cheers -ben cheers -ben

4 4

60445, empty version info stamps
by Ben Coman March 17, 2017

March 17, 2017

In 60445, browse the implementors of #cacheAt:for:ifAbsentPut: and the view the Versions of each. Curiously some of them have a blank user/date. There are others... errors := OrderedCollection new. versions := (CompiledMethod allInstances) flatCollect: [:cm| [VersionBrowser new setRGMethodFrom: cm asRingDefinition; buildChangeList] on: Error do:[:ex| errors add: (cm -> ex). #()] ]. versions size. "112131" emptyStamps := versions select: [:rgm| rgm stamp = '']. emptyStamps size. "9264" or 8% Maybe some are historic, but I'd guess this one at least is not so old... GLMFastListDataSource>>cacheAt:for:ifAbsentPut: What reasonable action do we need to take? cheers -ben

2 5

changing pharo packaging in linux
by Esteban Lorenzano March 17, 2017

March 17, 2017

Hi, Last weeks I was fighting with library dependency problems on linux distributions and at the end I came to the conclusion that the only way to guarantee that our own dependencies will be found first than OS versions (which can have problems for us when we choose to distribute a library by our own) is to add library directory to LD_LIBRARY_PATH before the execution. Now, I added that to zeroconf and thatâs working fine, but now we have the problem of the ânon zeroconfâ distributions. I checked how to add this and well, turns out that âold wayâ of distributions, from the standard builds of os-vm (we changed it like 5 years ago) already have something like that. The osvm for linux has this structure: ./pharo ./lib ./lib/5.0-201703221605 ./lib/5.0-201703221605/* //All binaries and libraries here ./bin ./bin/pharo thing is: here ./pharo is an script who mades all we need. So, instead reinventing the wheel I want to go back to this structureâ¦ I do not think it will imply anything to anyone, but I wanted to let you know :) cheers, Esteban

2 3

[pharo-project/pharo-core]
by GitHub March 17, 2017

March 17, 2017

Branch: refs/tags/60447 Home: https://github.com/pharo-project/pharo-core

1 0