On 05/21/2016 09:58 AM, WF Konynenberg wrote:
> Aye, from what I have read about these packed files here, it seems to
> me to be much like the various compressed file formats in use on
> Linux/Unix (.z, .gz, .bz, .xz, etc), except that perhaps the CMS pack
> tool neglects to modify the file name extension to indicate that this
> is now a packed file.  That means the "native" way to deal with these
> on Linux would be to implement some "cmspack" utility which can
> pack/unpack such files (as was indicated, the assembler source code is
> available so it's easy to figure out how to do this and reimplement it
> in a bit of C code), and teach the file command about this file format
> so it can correctly report it.

Filename extensions are not available to CMS (actual name of the
filesystem is EDF). But you were talking about something else, I
understand.


On 05/22/2016 01:12 AM, Alan Altmark wrote:
> There is no 'neglect'. Packed-ness is not tied to the filetype.

And that's a Good Thing. [tm]

Do one thing and do it well.


On 05/22/2016 07:22 PM, Mark Post wrote:
> Again, Alan's original suggestion was to teach cmsfs-fuse about such things.

Which I think is where Willy's heading. Sort of. (see below)


> I suppose one could use a combination of cmsfs-fuse and Rick Troth's 
> wonderful CMSFS package.

Thank you for the kind words, sir! I owe you yet another beer at
SCIDS.   :-)
Or maybe lunch at Chuy's in Austin ... something more substantial.

By the way, CMS FS (the older one) is now a Github project
<https://github.com/trothr/cmsfs> if anyone is interested. Contributions
welcome!
The FUSE based driver has surpassed it (can do read-write and actually
still works) but the older one has a utility mode which runs on any Unix.



On 05/22/2016 09:01 PM, WF Konynenberg wrote:
>   ...
> I'm just saying that any information that's available in the CMS file
> system "out of band" that is somehow relevant to an application trying
> to read/write any given file must be made available to the Linux
> programs in some way or other.   ...

Sure ... and RECFM and LRECL are available to CMS FS (either flavor) so
within a filesystem driver (or utility) a combination of RECFM=F and
LRECL=1024 with the magic numbers Alan posted would be strong indication
"this is in CMS Packed format".


>
> Cmsfs-fuse could provide this out-of-band info in various ways.

It could present a different Linux side filename. (with appropriate
extension tacked on)


> There's
> no "perfect" solution.   ...

Too true.

But something else:
Since all FUSE is user-land anyway, there's no reason why it should not
call out some user-land helper.
Maybe cms-fuse could call an unpacker, verbs like Hendrik mentioned.
(And then leave the fn.ft as-is.)

This violates "do one thing and do it well" (Mark alluded to,
consciously or not, saying filesystems typically don't take on these
tasks). But again, NNNN-FUSE is user-land. (We're not in Kernel anymore,
Toto.)
Just thinkin.


-- R; <><




----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390
----------------------------------------------------------------------
For more information on Linux on System z, visit
http://wiki.linuxvm.org/

Reply via email to