[darcs-users] Re: execute permission for scripts

Timothy Webster timw at outblaze.com
Thu Nov 25 21:01:58 UTC 2004

> > Following darcs manual, it would be great to enhance patch set of
> > addtional `file permission' patch type, but it will probably require some
> > research on how to represent cross-platform permission changes in the
> > best way.
> Yes, this is how it'll be done.  Except that I firmly feel that there is
> only one permission bit that should be versioned, and that is the execute
> bit.  The meanings of the other permissions are dependent on what computer
> the file is on, and thus shouldn't be versioned.

This may not seem related, but....

If people are trying to integrate real 'file permissions' then stuff like
acl's also should be considered.

So in my opinion the way to do this is to take a take a finger print
of 'file permissions' which can be applied as needed. For darcs to be
file system independent a finger print file should have a file system 
label. This of course the first person to set up ntfs vs ext3 extended 
acls vs etc needs to go through the through setting permission of 
files again. After that the proper 'file permissions' can be set each 
file system correctly.

I assume the default permission will be the current user permission 
as we do now and without execute bit, these are just files after all. 
Question should this be part of darcs or an externally called program
to create and apply 'file permission' finger print.

An external program would be real easy for me to create, first in perl
for ease and portability, then C later for speed required when dealing
with large repos.

> I presume that under windows execute permissions would just be ignored,
> which would be fine.  The only catch is that running optimize --checkpoint
> under windows would require keeping track of the permissions manually.

:( Sorry don't like the idea of "manual anything"


More information about the darcs-users mailing list