[Scummvm-devel] Subversion is here!

[Eugene Sandulenko]

On Thu, 19 Jan 2006 01:11:51 +0100
Max Horn <max at quendi.de> wrote:


> The question now is: Is there somebody out there who is willing to
> do that job? If yes, please tell me and also tell us a rough time
> frame. Otherwise, I still would like to go for a move during the
> coming weekend -- considering that no objections have been voiced yet
Well, I'd like to do the job, however. Dump file for scummvm-old and
scummvm modules (so merge would be possible) take 1.2G of plaintext,
and 16665 revisions (Jan 17th). Eek. Only way to handle that file
would be to write a simple program which will take such input:

  OFFSET line

That is, open the dump file in a hex editor which could show hex
offsets and text, biew is a good example. Search strings like Marcus
pointed out, write down offset of that line, and attach string which
will be inserted.

i.e.

  1000abcd Node-copyfrom-rev: YYY\nNode-copyfrom-path: old/file/path.cpp

Then a simple tool in perl would open the dump and that file with
instuctions, and combine them.

I could at least start this, though I am not sure I'll have enough time
to do whole task pretty fast. Then I would send file with instructions
to someone with better connectivity and he will construct a repository
from it which could be exported to sf.net then.

Ah, just got a better idea (IMHO). The dump file is so huge because it
contains full copies of each file revision. So another tool could throw
away those, and leave only dump-related keywords, prefixing them with
hex offsets. The job would be just to insert new lines into that. Then
new line will be inserted. Much much easier and more reliable. So I
think the task is more easier now. I could try to work on that these
weekends, and maybe couple hours in the next week, so work could be
done before release (and new tag is inserted, so screwing all offsets).


Eugene