On Tuesday 03 December 2002 3:01 pm, Walter Murray wrote:
> "Peter Smithson" wrote:
> > I'm told that you can not delete from "standard" record files yet I
> > have some documentation (quoted) which says you can. I find this
> > confusing!
>
> Peter,
>
> It would help if you could provide a little more information about the
> document you are quoting. Reading the quoted material in context might
> help us understand and explain the apparent contradiction.
Walter: I may be partly to blame for that conclusion, but that is because my
response was based on a "nit" -- his original question (and this followup)
talks about "standard" files, but his question on FDELETE actually applies to
RIO, or Relative I/O, files -- RIO != standard, hence the confusion.
[I've been meaning to point out this "nit" for a couple of days, but didn't
want to be accused of being overly nit-picky ;) ] Seriously, though, I'm of
the persuasion that, put simply, a "standard" file is one in which most, if
not all, of the bits in the FOPTIONS/AOPTIONS word are zeros (the bits that
control read vs. write/append; temp vs. perm; and so on don't count) -- HP's
documentation, however, actually does have a definition for a "standard"
file. See http://docs.hp.com/cgi-bin/doc3k/B3265090821.13550/65
2:3 File type option
Indicates internal record structure used to access
records in a file. If the file is old, this option is
ignored. Specifying a designator option (foption bits
(10:3)) value other than zero overrides this option.
This option is applicable only at file creation.
The following bit settings are valid:
000 Standard (STD) file
001 KSAM/3000 file
010 Relative I/O (RIO) file
011 KSAM XL file
100 Circular (CIR) file
101 NM spoolfile
110 Message (MSG) file
Default: 000
My definition would extend to things like fixed vs. variable/undefined; (or
even byte-stream) and ascii vs. binary, but I wouldn't quibble much over
someone claiming that a undefined--record-length binary file is "standard".
* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *
|