Re: [ gnuplot-Patches-1105717 ] Add '=' special file to reread '-' inline data

[Ethan M. <merritt@u.washington.edu>]

>=20
> >Comment By: Sampo Kellom=E4ki (sampo2)
> Date: 2005-01-20 19:12
>=20
> Message:
> Logged In: YES=20
> user_id=3D1200825
>=20
> BTW, I would much prefer to reply by email and you would get
> quicker replies. However, the message I got from you by mail did
> not have usable reply address. This is bad netiquette in my
> opinion.

You'll have to take that up with SourceForge.
The messages you are reading are auto-generated by the SourceForge
site itself whenever a comment or file is uploaded to
the website.  I agree that it is annoying, and they should=20
auto-generate some real return address.  But they don't.

If you want to join us in Email discussions, please post
and subscrive to the mailing list
	<gnu...@li...>


> I'd say that=20
> a proper redesign using temporary file for pipes might be
> the best thing to do.

That was the topic of discussion on the mailing list recently.
A temporary file is probably the most general answer.
But if you just want to replot the *same* data without
having to read it in again, it should be possible to use
the copy already stored in gnuplot's internal data structures.
So in that sense there are two separate possibilities:

Could re-use existing data with no actual re-reading:
  plot "-" with lines, "=3D" with points
  replot

Never stored the data in the first place, so clearly
would need to re-read:
  plot "-" using 1:2, "=3D" using 3:4


> However, I offer the '=3D' construct as design level
> suggestion on how to do it.

OK.  But still, do you propose that "=3D" is only valid
within a single plot command, or would you also be able
to use it in successive plot commands?=20


> If you can confirm that '=3D' is the official way to do it, I
> can move on with my life

There can be no promise that this will become "the official way"
when we don't even have a fully-working example at the moment.
If there is general agreement that this is a good target syntax,
then it is likely that something of the sort will eventually
make it into the official source.  But the exact details clearly
won't be settled until it has been successfully tried out in
patches to the development tree.

> Until then I'll use my patch even if it has its limitations.

Sure.  Although if you are working on additions to gnuplot I think
it would make sense to base your work on the development tree source
rather than on 4.0.


>=20
> Comment By: Hans-Bernhard Broeker (broeker)
                                     ^^^^^^^

By the way, you can reconstruct real Email addresses by knowing
that the information above refers the username on SourceForge.
So to reply to Hans-Bernhard you could have used
	br...@us...





> Date: 2005-01-20 11:44
>=20
> Message:
> Logged In: YES=20
> user_id=3D27517
>=20
> Error returns from ftell() definitely have to be checked
> here.  In particular, you have to be prepared to int_error()
> out of this if '-' is not a seekable file, i.e. if gnuplot
> is reading from a (redirected or console) stdin instead of a
> script.
>=20
>=20
> ----------------------------------------------------------------------
>=20
> Comment By: Ethan Merritt (sfeam)
> Date: 2005-01-20 05:48
>=20
> Message:
> Logged In: YES=20
> user_id=3D235620
>=20
> Very interesting.
> And either a remarkable coincidence or pretty quick
> programming, since we were just the other day grousing about
> the lack of such a feature.
>=20
> I've attached the equivalent patch against current CVS.
>=20
> Don't you have to test for ftell() returning EBADF or EINVAL?
> Is it  across multiple plots? E.g.:
>    plot '-' with lines
>    ... do some stuff...
>    plot '=3D' with lines
>=20
> I tried that, and it put gnuplot into a very strange state.
> Every keystroke or mouse click caused a replot but I could
> not exit the program.  But I may have used too strange of a
> test case.
>=20
>=20
> ----------------------------------------------------------------------
>=20
> You can respond by visiting:=20
> https://sourceforge.net/tracker/?func=3Ddetail&atid=3D302055&aid=3D110571=
7&group_id=3D2055
>=20

=2D-=20
Ethan A Merritt       merritt@u.washington.edu
Biomolecular Structure Center
Mailstop 357742
University of Washington, Seattle, WA 98195