[Monotone-devel] "branch" Parameter in get_netsync_write_permit()?

[Logan Sackette]

Hi All,

First off, thanks for the product and all the greats
hands that guides it.

I have been using mtn for some time now. I stated with
0.14 and really got into it at 0.15.  I really like
mtn and I started serving collections with 0.17.

My question is this, why did you guys remove the
branch/collection parameter
get_netsync_write_permit()?  I can see the reason for
changing collection to branch for the *read*().  But I
think both read and write should take the same
parameters.

Here is why.  I have branching I don't mind making
read-only, and I can do that with the current *read*()
hook.  Additionally, I want to restrict certain ids to
certain branches (mainly because I am serving many
collections from the same database).  This is in part
due to the inflexibility of serving a different
collection on a different port.  That is just too much
hassle and harder to maintain in my case.  Even though
I only have a few branches.

My lua hooks for the read/write function parses a UNIX
like passwd file to determine who has read/write
permission to which branch.

The format of my file is:
branch/collection: user_spec, ....
user_spec = keyid(perm)
keyid = is in the form of an email
perm = one of r for read-only, w for read/write (since
write only doesn't make sense).

For example, i can have a file like this:
net.venge.monotone: all(r), address@hidden(w)
net.venge.monotone.win32:person3(w)
...

The way the script work, it is possible to grant a
person read access all branches of a collection, but
only-write access to certain branches of the
collection.

In short, because I would like to do something like
this, I would really appreciated if you guys would
consider adding the branch parameter to
get_netsync_write_permit().

Thanks.

.v



        
                
______________________________________________________
Click here to donate to the Hurricane Katrina relief effort.
http://store.yahoo.com/redcross-donate3/