On May 11, 9:25am, Jim Wowchuk wrote:
> Subject: Re: LISTF, access proposal
> I would just like to echo Elbert E Silbaugh's comments about the LISTF,access
> proposal.
>
> 1) An intrinsic could provide the means to get what ever data is desired,
> in a format most suitable for the application. This neither denies the
> benefit provided by third parties like VESOFT, nor demands an
> unsatisfactory parsing excercise. The HP3000 should not need to become
> AWK-ward! :)
Elbert, Stan, Evan and now Jim have suggested that a user-callable intrinsic
(probably new items to FLABELINFO (and maybe to FFILEINFO?) may be more
valuable to those who have asked for this functionality, than enhancing the
exiting LISTF[ILE] commands. I want to do what best meets the needs our
customers within my relatively small time allotment, and I believed that adding
accessor info to LISTF[ILE] was the best way. There have been enough comments
from respected 3000 experts that I am open to changing my mind.
If CSY only provides an intrinsic then you will have the following choices: 1)
write a program that calls the intrinsic, 2) buy a program that calls the
intrinsic (may need to buy support too?), 3) get the program free from the CSL
(may not be an Interex member?, support?), 4) get a program free from somewhere
else (maybe even jazz, support?). Are these *preferable* alternatives?
Remember, it you want to automate some task that involves knowing a file's
accessors and you *don't* have this intrinsic, then you will have to write
scripts that probably will parse the output of :listf,access. This means that
CSY is not free to change/improve the output without possibly breaking your
script.
There is not enough time to do both an intrinsic and a command that calls this
intrinsic.
>
> 2) If a CI command is still desirable, then provide two -
> a basic one line/accessor abbreviated version and a full, whole
> kit'n'caboodle, verbose version.
>
> In considering the former, I can recall only two situations that occur
> where I need this information: a) I'm trying to *eliminate* all
accessors
> to a file; or b) I'm trying to get another program to concurrently access
> a file used by others.
>
> For the first situation I need information to uniquely identify the user
-
> a session number, pin number, and possibly the connection device. After
> all, I would prefer to ask them to exit the program than simply abort
> them, so anything that can identify the physical user becomes helpful.
>
> In the second situation, I need to know the flags for access - SHR/EAR,
> LOCK, etc.
>
> In the case of the verbose version, having a wealth of information can
> be useful, but an overflowing gutful is just too much. So for the
> verbose version I believe it would be desirable if this information could
> be limited not in its content but its quantity. So I reckon you would
> need the ability limit this to a specific user, session, or pin.
Thanks Jim. These scenerios really help us make informed trade-offs in the
design of this ehnancement.
>
> But while we're on the subject, how about a way to do a LISTF on files in
> the NEW domain? :)
How would this info be used? What do you do today? Any work arounds? More
scenerios?!!
> ----
Thanks,
Jeff Vance, MPE Lab
>
> But while we're on the subject, how about a way to do a LISTF on files in
> the NEW domain? :)
> ----
> Jim "seMPEr" Wowchuk Internet: [log in to unmask]
> Vanguard Computer Services Compu$erve: 100036,106
> _--_|\ Post: PO Box 18, North Ryde, NSW 2113
> / \ Phone: +61 (2) 888-9688
> \.--.__/ <---Sydney NSW Fax: +61 (2) 888-3056
> v Australia
>-- End of excerpt from Jim Wowchuk
--
Jeff Vance
|