vfat

[marc gonzalez-carnicer]

hi, thanks for the reply

2007/6/7, Aaron Bentley <aaron.bentley at utoronto.ca>:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> marc gonzalez-carnicer wrote:
> > stored on a pda). I keep a repository at my computer
> > at work, which runs windows. Since (afaik) it is not
> > possible to mount ext2 (or ext3, ...) partitions under
>
> Actually, NTFS support on Linux is quite decent now.  (Admittedly, it's
> a FUSE filesystem, but it seems to handle writes safely.)

yeah, trying to use ntfs could be a good idea


> This is a problem with vfat-on-linux, where the execute bit is assumed
> to be meaningful.  Under Windows, we know it's not meaningful, so no
> problem.

fyi, i run bzr on windows through cygwin, so it feels
like it's linux. and therefore the execute bit is significative


> Just to be clear: You have no ext2 partitions on this machine at all?
> It all uses VFAT?  Or is it just that your pen drive uses VFAT and
> you're doing the merge on your pen drive itself?

i use linux on all my personal machines, but at
work i *have to* use a windows xp machine.

i chose to partition my pendrive with vfat and ext2
partitions, but the windows machine can only mount
vfat. cygwin can not mount any partition that is not
vfat of ntfs. perhaps i should try having a ntfs on my
pendrive. do you think it may work?

i have to use the vfat partition of my pendrive to sync /
merge all my repositories. i used to use a remote free
subversion server but i switched to bazaar so that i could
sync in case of lack of internet access.



> If you have an ext2/3 partition somewhere on the machine, you can create
> a checkout of your pen drive and do the merging there:
>
> "bzr checkout /media/disk/path/to/branch /path/to/ext2/partition"

i assume you did not understand my scenario and
thought i only had linux machines ... :)



> > This page on the bazaar web site says vfat is not
> > appropiate for use with bzr. But then, what can i do?
> > Is there a hack i can do?
>
> It's not that FAT32 is completely unusable with bzr, but support for it
> is not something we've focused on.  However, using VFAT under Linux is
> not going to work well until we can detect that the execute bit is
> meaningless.

perhaps a solution would be adding an option to
bzr init so that the execute bit is not taken into
account for status, diff, merge, etc


> I tried some quick hacks, but they didn't work.

me too :)