> We  already  have  an  abstract  elisp  tree  reprensenting  archives,
> categories,  branches, versions,  and even  revisions. The  problem is
> that we allways refresh this tree. With a few lines of lisp, it should
> be possible to update the tree only when needed. 

I'll move completion engine and the abstract elisp tree representation
to xtla-core.el. I'll use your new process handlers and support revlib.
After that, please add the few lines.

> The problem with  [ar]browse is that if you want  to browse an archive
> with two categories, say A and B, and if B has a lot of branches, this
> will slow you down even while browsing A.

... yes, you are right.

[EMAIL PROTECTED] jet]$ /usr/bin/time -f %e tla abrowse -A [EMAIL PROTECTED] > 
/dev/null
55.09

[EMAIL PROTECTED] jet]$ /usr/bin/time -f %e tla categories -A [EMAIL PROTECTED] 
> /dev/null
6.22

Sigh. At the first time I guess timeof(network overhead) >> timeof(tla process
execution).

BTW, --snap option is interesting. Is there any good place in xtla to use 
a snap file effectively?

  

Reply via email to