Squeak-dev March 2007

squeak-dev@lists.squeakfoundation.org

143 participants
204 discussions

Summer of Code: 3 days left for the student applications
by Giovanni Corriga 23 Mar '07

23 Mar '07

Hi all, this is a friendly reminder to all the students interested in applying for the Google Summer of Code program. Remember that the deadline is Monday 26th (actually, Tuesday 27th at 00:00 UTC). This gives you a little more than 3 days for submitting your application, so if you haven't submitted one yet, you should hurry up! Giovanni

1 0

WriteStream>>contents
by Damien Cassou 23 Mar '07

23 Mar '07

Do you think that the following behavior is desired? stream := WriteStream with: 'something'. stream contents. => 'something' stream position: 2. stream contents. => 'so' We get the same result on Squeak and VisualWorks. Based on ANSI, #contents must return 'something' on both cases. -- Damien Cassou

3 2

Reimplementing WeakKeyDictionary
by "Martin v. Löwis" 23 Mar '07

23 Mar '07

I now have completed a first version of a reimplementation of WeakKeyDictionary, with the objective of improving performance (both computation and memory consumption). The entire change with a discussion can be seen at http://bugs.squeak.org/view.php?id=6348 In essence, this introduces two changes: - when values are finalized or keys removed from the dictionary, it is not always rehashed anymore. Instead, the dictionary just releases the value, and marks the association as expired. Expired associations can then be reused if a new value is to be inserted under the same has. Rehashing is only done when so many entries have expired. Expired entries don't count towards the size of the dictionary, and aren't reported when iterating over the dictionary. - WeakKeyAssociation was rewritten to be a weak class, with a fixed slot for the value, and a weak slot for the key (currently, it has two fixed slots, with the key slot referring to a WeakArray of length 1). In some measurements involving large (10000 keys) or many (10000) small weak key dictionaries, this new implementation improves performance by a factor of 4..5, when measuring finalization time. The second change saves 8 bytes per association (on a 32-bit machine). Regards, Martin

14 36

Howto change squeak bugs category?
by nicolas cellier 23 Mar '07

23 Mar '07

Hello, i incorrectly categorized some bugs in Kernel while they belong to Collection http://bugs.squeak.org/view.php?id=6366 http://bugs.squeak.org/view.php?id=6367 How to correct this? Nicolas

2 1

Beginning with TraitsOmniBrowser...
by Elod Kironsky 23 Mar '07

23 Mar '07

Hi! I just downloaded and started Damien's squeak-dev-93. Gratz for that Damien, great work! Currently I'm working on a client API for SmallDEVS and we decided to give Traits a tryout. For that I need TraitsOmniBrowser. I dowloaded that from www.squeaksource.com and loaded into the image. Everything went fine until now, but when I try to execute OBTraitStructureBrowser open. I get an MessageNotUnderstood in OBTraitStructureBrowser class >> addTo:class:comment:metaclass:. It seems that the classes OBCategoryActor and OBClassActor are not present in the system. My question is, where can I find these classes? Elod

2 4

Possible problem with FFI in 3.9-7067 ?
by Andrew Tween 23 Mar '07

23 Mar '07

Hi, I have been trying to use FFI (the latest on SqueakMap) in 3.9-7067, on Windows 2000, to get a device context, and do stuff with it. It doesn't seem to work like it used to. Some investigation reveals that everything worked ok up to 7052. For example, In 7032 - "Win32Window getFocus getDC getHandle" answers an integer In 7067 - "Win32Window getFocus getDC getHandle" answers a ByteArray() Inspecting the Win32Window>>#apiGetDC: compiledMethod in 7032 and 7067 reveals this difference... In 7032, literal4 = #Win32Window In 7067, literal4 = #Win32Window->Win32Window recompiling the method in both images makes no difference. Also, in 7032 - "Win32Window win32Draw" works correctly, But in 7067 - "Win32Window win32Draw" crashes the VM (both the old, and the new VM) Have I installed the wrong FFI packages for 3.9? Or is there a problem in FFI? Or maybe there is a problem in the Compiler? Cheers, Andy

5 12

Chris Muller has a new e-mail address
by Chris Muller 23 Mar '07

23 Mar '07

My old e-mail address has been having problems, my new address is: ma.chris.m (at) gmail.com Please send all future e-mail correspondence to this address. If you've sent something to my old address in the last month or so, I may not have received it. If you are obliged, I would be happy to read it at my new address. Thank you. - Chris Muller

1 0

Chris Muller has a new e-mail address
by Chris Muller 23 Mar '07

23 Mar '07

Greetings, my old e-mail address has been having problems, my new address is: ma.chris.m (at) gmail.com The "(at)" above means the @ character. That confusing notation above is just my attempt to minimize spam to my new address. Please send all future e-mail correspondence to this address. If you sent something to my old address in the last month or so, I may not have received it. I would be happy if you resend it to my new address. Thank you. - Chris Muller

1 0

3.10alpha: PluggableText in SqueakMap showing boxes
by trilobyte＠hoe.nu 22 Mar '07

22 Mar '07

Hi, The PluggableTextMorph in SqueakMap under 3.10alpha shows hollow boxes in place of cr characters. I searched Mantis but couldn't find anything. Anybody have any clues? Thanks, Tim

3 4

NuBlt Update
by Eddie Cottongim 22 Mar '07

22 Mar '07

I thought I'd give an update on my BitBlt reimplementation in case anyone was interested for SoC or otherwise. Since I last mentioned it here, I've moved it to Slang, so it is developed and built similarly to the regular BitBlt. Large portions are auto-generated from templates (namely many of the combination rules and pixel depth conversion), cutting down the amount of real code to be developed. While it can't yet run a Squeak image (well, not for long...), it can run some basic benchmarks and tests on itself. There are bugs, but it will at least run most of the internal benchmarks. The performance is promising, with most supported operations being faster than stock BitBlt, and a few being slower (these need to be special-cased). Some operations will not be supported soon, especially the WarpBlt functionality. I'm no longer working heavily on this, but do occasionally open it up to fix bugs, add docs, etc. The latest dump is available at: http://home.earthlink.net/~cottongim/NuBltMar21-2007.st.gz Eddie

2 2

← Newer
1
2
3
4
5
6
7
8
9
...
21
Older →

Jump to page: