Closed GoogleCodeExporter closed 9 years ago
I think, if SMB_URL_LEN is not protocol-specific limitation, we should also
treat the
files which are not feet in buffer (at smbwk.c, line 151). Also, some compilers
(i.e.
msvs), has sizeof(off_t)==4. May be, fpos_t is more sufficient type for file
size.
Original comment by radist...@gmail.com
on 31 Dec 2009 at 10:06
off_t is type of seze item in the struct returned by smbc_stat(). If you want
something else, then I suggest long long.
Original comment by ruslan.savchenko
on 31 Dec 2009 at 10:45
3. In smbwk.c change SMBSCAN_GO into SMBWK_GO
Original comment by ruslan.savchenko
on 31 Dec 2009 at 10:46
>I think, if SMB_URL_LEN is not protocol-specific limitation, we should also
treat the
files which are not feet in buffer (at smbwk.c, line 151)
AFAIK, passing an entire url is the only way to access a file with
libsmbclient. We
can always enlarge the buffer, if it appears to be small for our needs. Too long
paths/filenames are not good at all (recursive links for example).
Original comment by ruslan.savchenko
on 31 Dec 2009 at 10:51
4. More inforamtive messages on smbws_url_append fails
Original comment by ruslan.savchenko
on 31 Dec 2009 at 10:59
> off_t is type of seze item in the struct returned by smbc_stat()
We should somehow handle the large files.
> We can always enlarge the buffer, if it appears to be small for our needs.
Don't need it. Just alloca-ed buffer if the case of failed smbwk_url_append
call at
line 151.
Original comment by radist...@gmail.com
on 31 Dec 2009 at 11:03
>> off_t is type of seze item in the struct returned by smbc_stat()
>We should somehow handle the large files.
Type returned by library won't change. But in dt_dentry size _must_ be of 64bit
type
because entire directory size is accumulated in it.
>> We can always enlarge the buffer, if it appears to be small for our needs.
>Don't need it. Just alloca-ed buffer if the case of failed smbwk_url_append
call at
line 151.
Now i see the point
Original comment by ruslan.savchenko
on 31 Dec 2009 at 11:10
5. Support for large file names. Scanner can ommit scanning a directory if
resulting
path is too long, but it should scan all the files in accessed directories.
Original comment by ruslan.savchenko
on 31 Dec 2009 at 11:12
6. long long type for dt_dentry.size
7. Replace two tree walks in dt_mktree by one (if possible)
Original comment by ruslan.savchenko
on 31 Dec 2009 at 11:15
8. SMBSCAN_LOG macro for error messages
Original comment by ruslan.savchenko
on 31 Dec 2009 at 11:35
9. Simplified output:
first line:
<total number of directories>
next lines have format:
<path_id> name <0 if file, id if directory> size
where path_id is id of parent directory
such format should allow easier parsing for high-level scanner
Original comment by ruslan.savchenko
on 31 Dec 2009 at 12:15
><path_id> name <0 if file, id if directory> size
<path_id> <0 if file, id if directory> size name
name can contain spaces. We do not want to encode names here...
Original comment by ruslan.savchenko
on 31 Dec 2009 at 12:16
10. Do not call smbc_open() when going up in file tree. This will cause
redundant
calls to smbc_close() - remove error checking after calling it.
Original comment by ruslan.savchenko
on 31 Dec 2009 at 2:19
11. print total share size (right now root directory is not printed)
Original comment by ruslan.savchenko
on 1 Jan 2010 at 2:12
topic dammit
Ideas 1,2,3,4,6,7,8,10,11 commited. 5,9 remain
Original comment by ruslan.savchenko
on 1 Jan 2010 at 2:18
5,9 commited
Original comment by ruslan.savchenko
on 2 Jan 2010 at 12:56
Original issue reported on code.google.com by
ruslan.savchenko
on 31 Dec 2009 at 9:20