Re: QTParted/reiserfs trashes

[Yury V. Umanets]

On Wed, 2004-03-03 at 18:21, Szakacsits Szabolcs wrote:
> On Wed, 3 Mar 2004, Yury V. Umanets wrote:
> 
> > reiserfsprogs is official progs from NAMESYS. And progsreierfs is not
> > official different implementation. 
> 
> Thanks for the explanation. So it seems that reiserfs has changed at some
> point, then reiserfsprogs were adapted/fixed but not progsreiserfs.

I'm sure, that reiserfs is not changed since that times. The reason of
bugs is possible bug in resize code of progsreiserfs or/and using it on
slightly corrupted filesystem. 

> 
> > It has fully different interface. Parted uses progsreiserfs, because
> > it is easy to use it. It has convenient interface and reiserfsprogs
> > has no.
> 
> Usually this is solved by refactoring the code and/or making wrappers thus
> no double effort. But probably I'm missing something ...
It is possible, but fact is that implement it from the scratch with
fully new interface was easier task thank adopt existent reiserfsprogs.

reiserfsprogs did not try to satisfy needs of all applications but fsck
and mkfs. So, it consists of pile of fully not structured code. There
are no abstraction levels, pseudo objects (like FS, superblock, etc.),
naming system, device abstraction, etc..

Thus, for instance, to make a filesystem, one should have to type 20Kb
of code or run external mkfs binary. The same and even more for other
tasks (copying, resizing, etc)

progsreiserfs has not ideal interface, but to create reiserfs filesystem
on some device it is enough to type something like this (pseudo code):

/* open device abstraction, this can use not only files */
dal_t dal = dal_open("/tmp/image", RW);

reiserfs_fs_t fs = reiserfs_fs_create(dal, FORMAT_3_6, blocks_nr, etc);
reiserfs_fs_sync(fs);

/* any actions (copy, tune, resize, etc) on opened/created fs though the
fs instance @fs. */


reiserfs_fs_close(fs);


dal_close(dal);


> 
>       Szaka
-- 
umka