Re: reading variably sized floats
- From: David Thompson <dave.thompson2@xxxxxxxxxxx>
- Date: Thu, 05 Feb 2009 08:56:50 GMT
On Fri, 23 Jan 2009 08:55:50 +0000 (GMT), nmm1@xxxxxxxxx wrote:
In article <0606c19d-d48b-4d4b-9d95-7e03d9781a87@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,I assume you mean the former (just data) not the latter (expand
Terence <tbwright@xxxxxxxxx> wrote:
Richard Maine wrote:
The tab character is just a character and is counted as 1 when using
the nX ( or the nA(m) or An) format field specifier.
While that is a likely behavior with most compilers, possibly even all
of the current ones, it is *NOT* behavior specified by the standard. The
standard does not require that tabs be accepted as formatted input data
at all. It would certainly be valid for a compiler to do something like
expand the tabs into equivalent blanks. I don't know that any current
compilers actually do this, but the standard does allow it.
Even more interestingly, that behaviour is precisely what is required
by the ASCII standard - but few people now realise that it is more
than a character set!
And, yes, I remember that happening, too.
spaces). It's been a long time since I had an actual X3.4, but my
strong recollection is that all the Format Effectors were defined only
for print/display (output) devices, and only in terms of moving the
active position. In particular, for many (I would say most) early CRT
terminals, sending space(s) would blank out any previous data in the
positions moved over, but HT would not. (Sometimes this was selectable
by a switch, jumper, etc.)
I don't recall any specific requirement that other equipment leave it
unchanged, although there's an argument that it's _intended_ for the
last device and thus everything that is -- or thinks it might be --
upstream has a responsibility to pass it on.
<snip>
The separate but related X3.64 did define (among many other things) aIn response to Richard's note, I do point out that any future action
on a Fortran-compiled executable, with respect to the postulated
automatic expansion of the tab character to some number of blanks,
(and I assume only in a "formatted input stream"), would require this
number of replacements blanks, or the definition of tab points, to be
User-declarable, to prevent gibberish.
Well, that's not what the ASCII standard says, not what used to happen,
and not what is conventional nowadays. For at least the past 20 years,
the de facto meaning of TAB has been to move to the lowest 8n+1 position
that is at least 2 positions beyond the previous one.
command sequence to set horizontal tab stops, which I saw used a fair
bit. But I agree that next-mod-8 was the single most common usage, and
still is where it is output at all (not processed internally).
But it almost always has been system dependent, and not standardisedSystem(s) and/or device(s).
between systems.
<snip rest>
.
- Follow-Ups:
- Re: reading variably sized floats
- From: nmm1
- Re: reading variably sized floats
- Prev by Date: Re: STOP in Fortran 2003
- Next by Date: Re: Elegant way to define a callback interface
- Previous by thread: Where do I find group info?
- Next by thread: Re: reading variably sized floats
- Index(es):
Relevant Pages
|
