>
>
>
>No, not block, just reuse the already-opened environment. When BDB opens
>our FS environment, it mmaps the environment files. That's just fine on
>most systems, but HP-UX does not let you map the same file into a
>process twice. So my longish-term plan is have a reference-counted cache
>of open environments, and avoid the second (and third, and fourth...) mmap.
>
>
OH cool,
I got scared for a moment.
I guess I should mention again that I don't care for the separate
svn_fs_new followed by either a create or an open.
I have asked before why it is that way. I've gotten no response so far.
SOOOOOOO:
For the sake of argument, lets assume that the separate calls aren't
required and we kill off the svn_fs_new function.
Then, svn_fs_open _fs (notice name change :-), and svn_fs_create_fs
become simple factory methods that return FS objects.
svn_fs_open can determine at the time of calling whether the mapping has
taken place already. If so, return the old FS object.
This may cause some issues with pools but, but I think it can be worked out.
Also, I'm not clear why this needs to be on disk at all. When the
process goes away, so does the problem. Is this a windows thing?
gat
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Mar 17 22:16:25 2003