Re: Variable length/precision formats?

robin wrote:
....
Since VFEs use wierd, indirect, on-the-fly modification of the
format

No, they do NOT require modification of the format.

*while* the I/O list it controls is still being processed,
and since this fool is still promoting them,

The only fool here is YOU, because you fail to grasp
how this feature is implemented.

I have implemented the feature. I'm certain you have not.
The feature requires something like "thunks" that change
the format as the I/O list is processed. This has already
been fully discussed in this thread. The value of the
format changes every time one of the variables used
in any of the VFEs in the format string gets changed. This
includes changes made to those variables on-the-fly: *WHILE*
the I/O list is being processed. Variables such as implied
loop indices and I/O list items themselves can be in the
VFEs! This wierd, indirect, on-the-fly interaction is not
necessary or desirable. It is a common cause of error
among those using the feature. These errors could be
entirely eliminated by removing the wierd, indirect, on-
the-fly behavior of VFEs. Indeed,the only result of such
a change would be an increase in clarity of programs
using the feature.

Of course, there are other weaknesses of the fearture.
It can't be applied to character variables (or even named
constants) but only to literals inside I/O statements and
to the text of FORMAT statements. That limitation should
be removed. They can't be applied to strings used for
purposes other than I/O. That limitation should be removed.
With those useful changes, I would even support the putting
the feature into the standard. As it stands, I recomment users
never even read about the feature, much less use it. I've
said all this before.

....
]>> it's obviously the
case that he *wants* wierd, indirect, on-the-fly modification of the
format

You are lying.
There is no earthly reason that the values used
as VFEs cannot be established for the write prior to
interpreting the format specification.

Well, the "earthly reason" is that the specification of the
feature (non-standard but shared by most implementations)
*REQUIRES* that VFEs continue to be re-evaluated *WHILE*
the I/O is proceding. I think a behavior that's required by the
definition of the feature is an "earthly reason". I can't imagine
such a poor design in paradise. Fortunately the probability
the standard will ever include the fearture is exactly zero.
I recommend that future Fortran implementations skip it too.

--
J. Giles

"I conclude that there are two ways of constructing a software
design: One way is to make it so simple that there are obviously
no deficiencies and the other way is to make it so complicated
that there are no obvious deficiencies." -- C. A. R. Hoare


.

Relevant Pages

  • Re: Variable length/precision formats?
    ... having it work like VFEs ... indirect interactions with the I/O list, ... The feature is ... the *FORMAT* is still list-directed in the above. ...
    (comp.lang.fortran)
  • Re: Variable length/precision formats?
    ... they do NOT require modification of the format. ... I have implemented the feature. ... in any of the VFEs in the format string gets changed. ... the "earthly reason" is that the specification of the ...
    (comp.lang.fortran)
  • Re: Numeric literals in other than base 10 - was Annoying octal notation
    ... If you want your data file to have values entered in hex, or oct, or even ... able to choose to prepare the data in a certain format - the format ... When designing a language feature don't be too ready to ... impossible or unreasonable to support a feature then by all means ...
    (comp.lang.python)
  • Re: Question about blitting from RGB surface to YUV surface
    ... so if a spec doesnt ask for a feature, the hw guy wont think to add it, ... That and YUV is typically seen as a video format and not a 3D format. ... conversion needed, I should do the work programmly. ... PC with ATI graphic card, I notice that a RGB-format overlay surface was ...
    (microsoft.public.win32.programmer.directx.graphics)
  • Re: Excel can automatically put in formulas...or not?!
    ... It is called Extend data range format and formulas, ... In some circumstances, when I type input into a column in a new row, Excel ... Does this feature have a name that I could use to look-up in Help? ...
    (microsoft.public.excel.misc)