On 12/04/2011 06:51 PM, Richard W.M. Jones wrote:
On Sun, Dec 04, 2011 at 11:18:37AM +0800, Wanlong Gao wrote:
> Hi Rich:
>
>>
>> ACKed and pushed,
>>
>> Thanks!
>>
>> Rich.
>>
>
>
> I saw that you killed the '-p' option because RHEL5 didn't support this,
> I want to know that which version of blkid you use? or which util-linux?
> or which RHEL5 updated release?
>
> Because in my RHEL5.7GA, the '-i' option isn't supported, either.
Yup, you're right ...
For '-p': '-c /dev/null' should be completely equivalent and it's
what
we use in the rest of daemon/blkid.c. Is there any other benefit to
using '-p' instead of '-c /dev/null'?
I think no, the '-p' option can also returns the info of 'partition
table type (PTTYPE) and partitions (PART_*). although the cache func
likes '-c'.
Obviously I didn't test this, which is why I didn't see that the '-i'
option isn't supported on RHEL 5 ...
> But does the '-i' option supported on RHEL5.8?
I'll need to add a test for it. It seems pretty straightforward just
comparing the blkid --help output on RHEL 5:
$ /sbin/blkid --help
/sbin/blkid: invalid option -- -
blkid 1.0.0 (12-Feb-2003)
usage: blkid [-c <file>] [-ghl] [-o format] [-s <tag>] [-t
<token>]
[-v] [-w <file>] [dev ...]
-c cache file (default: /etc/blkid/blkid.tab, /dev/null = none)
-g garbage collect the blkid cache
-h print this usage message and exit
-s show specified tag(s) (default show all tags)
-t find device with a specific token (NAME=value pair)
-l lookup the the first device with arguments specified by -t
-v print version and exit
-w write cache to different file (/dev/null = no write)
dev specify device(s) to probe (default: all devices)
versus upstream:
$ blkid --help
blkid: invalid option -- '-'
blkid from util-linux 2.20.1 (libblkid 2.20.0, 19-Oct-2011)
Usage:
[...]
-p low-level superblocks probing (bypass cache)
-i gather information about I/O limits
It is sufficient to grep the blkid --help stderr for '-i'.
> I saw that the latest e2fsprogs also don't support the '-p' and
'-i'
> options, So I'm considering add these options into e2fsprogs, is it
> needed?
Which e2fsprogs program do you mean?
The *libblkid* and the command *blkid* (the interface of 'libblkid') is
developed in e2fsprogs(which is the user tools of ext2/3/4fs), but it
seems that the *libblkid* and command *blkid* is separated from the
*e2fsprogs* project and go into the *util-linux* project with the supporting
to others filesystems more well and adding a lot of features like
*partitions* and so on, and the *-p* and *-i* options is supported.
I think using the *-p* and *-i* options to return the low-level of device
status is the very useful features, but it isn't supported on RHEL5,
because the RHEL5 is too old for us.
So, for the future of *libguestfs*, can we add something to separate these disto?
For example we can remove the features like *-p* from RHEL5 but add it to RHEL6 or
some fedora distos?
Can we just FEATURE this distos?
Or:
remove the options like *-p* in *blkid* and so on or just split it from the *stable*
release of *libguestfs*, but the I think the later is a patient thing.
Is it time for us to consider for these things now?
Rich.