July 2009 - Pharo-dev - lists.pharo.org

[Pharo-project] Double draw??
by Schwab,Wilhelm K July 20, 2009

July 20, 2009

Hello all, I saw Damien's responses to my report of #localHostName troubles, and began a little fact checking for a follow up. That lead me to some browsing, in particular to senders of #primHostNameSize. I am running on Linux on an old machine and with an image having many of the same changes as in the recent "fast" image, but it's still not as snappy as one would want. That said, #testIPv6Echo is a long method, and when I select it in the senders list, I can see the code render in plain text and then a second time syntax highlighted. It makes me suspect that every method is rendered twice, and there might be a speed boost if we could go directly to the plain or colored source in one shot. I do not notice the double-draw with shorter methods. Note that the double-draw could be unique to Linux, possibly contributing to its sluggishness??? Bill

2 1

[Pharo-project] Removing SmallDictionary freezes the image
by StÃ©phane Ducasse July 20, 2009

July 20, 2009

Hi all Cyrille cleaned SmallDictionary and check tests (169 now) Now he wanted to remove SmallDictionary before loading his new version and when we do that the image simply freezes.... And there is no references beside a tests there are no instances. So either the reference finder is broken or I do not get it. Does anybody has an idea what could be problem? Philippe I remember that during the sprint you got a strange strange behavior with SmallDictionary already and that we had to split that into chunks. update10365 "self new update10365" self withUpdateLog:'- Step 1 of 3 (change superclass) - Issue 463: SmallDictionary should not be a subclass of Set (and Dictionary) - http://code.google.com/p/pharo/issues/detail?id=463'. self script443. self flushCaches. self withUpdateLog:'- Step 2 of 3 (change methods) - Issue 463: SmallDictionary should not be a subclass of Set (and Dictionary) - http://code.google.com/p/pharo/issues/detail?id=463'. self script444. self flushCaches. self withUpdateLog:'- Step 3 of 3 (remove methods) - Issue 463: SmallDictionary should not be a subclass of Set (and Dictionary) - http://code.google.com/p/pharo/issues/detail?id=463'. self script445. self flushCaches. Stef

3 6

[Pharo-project] need tests for DataStream
by Cyrille Delaunay July 20, 2009

July 20, 2009

Hi, I wrote a new version of DataStream ( and its subclasses : ReferenceStream and SmartRefStream) in Nile and I would like to write tests for those classes. I would be very interested by anyone that has already write some tests for that.

2 1

[Pharo-project] BlockArgsNode KeyWordNode
by StÃ©phane Ducasse July 20, 2009

July 20, 2009

Hi I'm wondering why these classes are the only one of Compiler-Syntax and not Compiler-ParseNodes. And why nobody use them. Stef

3 3

[Pharo-project] How to have a faster image
by Mariano Martinez Peck July 20, 2009

July 20, 2009

Hi! I want to put together the things people do to have a faster Pharo. The "slow" stuff is being identified and for the moment, perhaps it is useful people can have an image with less features but faster. All the things we put here, perhaps we can then include them in the script used to generate Pharo until the things that now are slow, are fast. This is what I do when I download a Pharo image to get it faster: ToolSet default: StandardToolSet. SystemBrowser default: OBSystemBrowserAdaptor. Preferences enable: #fastDragWindowForMorphic. Preferences disable: #dynamicProtocolActivation. Of course, there are other things like using the default Squeak theme or not using FreeType but in my case I rather use them. I hope this helps to someone. Best, Mariano

6 15

[Pharo-project] About issue number in slice name
by StÃ©phane Ducasse July 20, 2009

July 20, 2009

Hi all I would like that we follow the convention: - put the issue number in the slice name (like nicolas is doing) Today the server crashed and the inbox is full of all code. So with the name in the slice I can really fast check if it was integrated or not. So we should probably update the wiki with that wish (important for me). Stef

1 0

[Pharo-project] due to the squeaksource crash... the inbox is full of old
by StÃ©phane Ducasse July 20, 2009

July 20, 2009

versions I moved in the treated inbox during the sprint.... So I will clean again. May be we should pay somebody to improve the infrastructure. Stef

1 0

[Pharo-project] New much faster Pharo based on 10379
by Damien Cassou July 20, 2009

July 20, 2009

I've just uploaded a new Pharo image. It is much faster now, but you won't get the package browsing (only the category browsing). http://pharo-project.org/pharo-download For all the people who found Pharo slow, please tell me what you think of this one ! -- Damien Cassou http://damiencassou.seasidehosting.st "Lambdas are relegated to relative obscurity until Java makes them popular by not having them." James Iry

8 13

[Pharo-project] NetNameResolver
by Schwab,Wilhelm K July 20, 2009

July 20, 2009

Hello all, I noted some problems with NetNameResolver, and found the following: http://n2.nabble.com/NetNameResolver-problem-on-Linux-Ubuntu-9.04-td2805740… However, there are problems on Windows too - in fact, I suspect the following would be a defect on any platform: #localHostName uses #primHostNameSize (ipv6 specific, right??) "unprotected." Looking at the other methods, it should check #useOldNetwork, though I am confused about how it should act based on the result. Any suggestions? I have also managed to create (aka steal from 3.9) a working #nameForAddress:timeout: method, which IIRC, was a disaster the last time I tried it on Pharo. I was running it on win2k3, and should try it on Ubuntu to see if that was the source of the trouble. Bill

2 2

[Pharo-project] Problematic method category
by Schwab,Wilhelm K July 19, 2009

July 19, 2009

Hello all, I have noted that some of my imported code has method in a category named **auto-generated**, and Monticello/PackageInfo sees it as a package. Is there a way to browse all methods in a named category, or better yet a sweeping progammatic fix to change the category name of all affected methods? Bill

2 1