Personal info for maf

This person is currently certified at Apprentice level.

Name: Markus Fritsche

Homepage: http://reauktion.de/archer/

Notes: I'm interested in

This person is:

Recent diary entries for maf:
RSS

5 Apr 2005  »

A small squeak...

Well, I've got a new server (actually, it's a "vServer", a UML-like linux). For now, squeak runs fine on it, serving a SmallWiki at http://archer.kicks-ass.org:8080/, but this one's (on top of Squeak3.8) needs 20MB of ram for it self. With a softlimit of 80MB of ram and a hard limit of 100MB, there's not much to experiment with - which reminds me of the various 'Small' squeak approaches which are on their way, but had not stepped the stage yet... Well, there's a QuiteSmall-Image, but it's 3.7 based and doesn't contain a number of hard-to-track-down methods and fixes you'll need for SmallWiki and stuff. Well, I'd like to see spoon progressing, but afaic, first it's not there yet and second, it depends on flow, which in turn, seems to be a replacement for the whole collection and (network-) streaming library (which would make it hard to adapt the rest to it - correct me if I'm wrong). And even if I try to use what I've learned of CompiledMethods, it would be quite a lot of work to build something which seeds a 'core' image with needed Behavior.

Maybe I should contribute a small amount of vServers to the people to build up a need for ram-saving squeak images ;)

14 Mar 2005 (updated 14 Mar 2005)  »

On my trip to CompiledMethod land, I learned a lot about them and still do. Also, I'm slowly becoming aware of how spoon is supposed to work. Right now, I know of twelve different kinds of literals:

With Association being: Reference to

And ReadOnlyVariableBinding beeing a reference to a class which was defined at compile time.

8 Mar 2005  »

On sunday, I started my worh on Knigge, a programm which should (if ready) be able to transfer Classes and CompiledMethods from on image to another, by files or over the network. I had to tweak Classloader a bit and it already shows a need for better test cases and a bit refactoring.
Until now, Knigge doesn't care much about dependencies - which leaves you with corrupted CompiledMethods sometimes (save early, save often). However, looking up dependencies (superclasses, literals of CompiledMethods) maybe should done a layer above Classloader.
I'm every time impressed how fast you get an idea of the 'deep inner basic' by looking at the code and using the Inspector here and there...

3 Mar 2005 (updated 3 Mar 2005)  »

Composing my fat development image, I wish we had some kind of "atomic fileIns". Because, sometimes, things break (ever seen the emergency evaluator?). The process would be: load the code, compile it, and if it compiles properly in whole, then swap the classes and methods. That wouldn't help in cases where the code is broken, but on the other hand, if you have an incomplete fileIn, it would help you not sitting with a half of new methods and the other half awaiting old behaviour... So, in the spirit of OpenSource, it's my task to get it done. Let's look if I can handle it ;)

This person has certified others as follows:

Others have certified this person as follows:

[ Certification disabled because you're not logged in. ]

[ Home | Articles | Login/Account | People | Projects ]