Hi,
I've been doing some testing:
1) I stuffed a block file full of 1.3 MB's worth of data, in 30 blocks and
it was still speedy.
2) I added 10 000 blocks to a block file. Around 5000 it took 2 seconds for
100 blocks, around 9000 this increased to about 3 secs
3) After adding 10 000 blocks the app took 1 MB of RAM
4) saving all these blocks to disk the first time took 4 seconds, yielding
a 365k file (guess how large each block's data was ???).
5) after saving the file, the application only took 201k
Does this sound good so far, or is this too slow? BTW, this was
hand-counted, not by the Mac, so it may be a bit faster actually. All of
this was done on a 90MHz PowerPC 603e (Performa 5300). The 680x0 version
should require less RAM as data structures are padded rather generously on
PowerPCs.
For 20 000 blocks I got 3.1 ticks per block, and saving took about 15
seconds, resulting in a 725k file. But this will certainly be faster in
real-life situations as then only 10 or so blocks will have to be written.
BTW, before Anthony starts yelling for a more effective look-up of blocks:
I plan to replace the current linear search of blocks in favor of a tree
structure in RAM to save a block hierarchy. This will result in "owned"
blocks, where one can search all blocks which are "childs" of a particular
block. E.g. this would then only look for buttons on one card, and not in
the whole stack as would be necessary right now.
Cheers,
-- M. Uli Kusterer
------------------------------------------------------------
http://www.weblayout.com/witness
'The Witnesses of TeachText are everywhere...'
--- HELP SAVE HYPERCARD: ---
Details at: http://www.hyperactivesw.com/SaveHC.html
Sign: http://www.giguere.uqam.ca/petition/hcpetition.html
