6:06 p.m.
Dear all,
I am sending this email because lately I have observed changes in the way gretl interacts
with octave. I am attaching a simple program for illustrative purposes. I think that such
a program should work in any case.
I think it works in linux with no problems (based on earlier experience with more detailed
programs to produce graphs).
The issue is that I was able to run this program under windows 7 with earlier gretl and
octave versions, in particular it works under gretl version up to 2016d and old octave
versions - works with octave 3.4.3. However, under win 7 it does not work with late octave
versions - does not work with octave 4.2.1.
In any case, I cannot make it work under win 10, no matter which version of gretl (I have
used 1.9.92, 2016d, 2017d) and octave (3.4.3, 4.2.1) I use.
Could you take a look on this issue?
Kind regards,
Andreas Zervas
P.S. I originally started coding with octave because gretl does not offer a facility to
produce subplots (graphpage is an imperfect substitute, and accepts only up to 8 graphs I
think, while scatters accepts only one variable on each graph). I also did not like the
idea of learning another language just to plot, when I am familiar with matlab / octave.
Any plans for such a feature?
6:19 p.m.
Am 25.05.2018 um 20:06 schrieb Andreas Zervas:
I think it works in linux with no problems (based on earlier
experience with more detailed programs to produce graphs).
Right.
The issue is that I was able to run this program under windows 7 with
earlier gretl and octave versions, in particular it works under gretl
version up to 2016d and old octave versions - works with octave 3.4.3.
However, under win 7 it does not work with late octave versions - does
not work with octave 4.2.1.
In any case, I cannot make it work under win 10, no matter which
version of gretl (I have used 1.9.92, 2016d, 2017d) and octave (3.4.3,
4.2.1) I use.
Could the standard paths have changed? So, what is your gretl
preference
setting for the path to octave? Is it correct? Can you run any octave
foreign block at all?
P.S. I originally started coding with octave because gretl does not
offer a facility to produce subplots (graphpage is an imperfect
substitute, and accepts only up to 8 graphs I think, while scatters
accepts only one variable on each graph). I also did not like the idea
of learning another language just to plot, when I am familiar with
matlab / octave. Any plans for such a feature?
Yes we recently had a discussion
about it on the devel list (thread
"integrating gnuplot's multiplot"), although not very deep. I guess it
will boil down to putting together a function package, I would be
optimistic that the ETA is this year, but more pessimistic about this
summer.
Maybe you can go into more detail about what kind of plot type you have
in mind.
thanks,
sven
3:18 p.m.
New subject: multiplots / subplots (was: Re: Bug?)
Am 25.05.2018 um 20:19 schrieb Sven Schreiber:
Am 25.05.2018 um 20:06 schrieb Andreas Zervas:
> P.S. I originally started coding with octave because gretl does
not
> offer a facility to produce subplots
will boil down to putting together a function package, I would be
optimistic that the ETA is this year, but more pessimistic about this
summer.
Well actually I thought today that I need this myself, so I'm attaching
a first attempt which so far works for me. (So far it's a plain
function, not a package.)
The first arguments for this function 'multiGplot' are these:
1) Array of strings with the details for the plotting command - so the
stuff that comes right behind the 'gnuplot' keyword.
2) A string with (double-dash) options that all subplots have in common.
3) A gretl list with all the series inputs that are referenced in the
individual gnuplot commands. (Can use shortcut 'deflist(dataset)' for
smaller datasets.)
4) Integer with the number of columns in the plot grid. (Optional,
default 1.)
5) A string with a file/path name to determine gnuplot's output, in
parallel to the display output. (Optional, default none/null.)
Examples (referring to the denmark.gdt data):
multiGplot(defarray("LRM --time-series --with-lines", \
"LRY --time-series --with-lines"), null, \
deflist(dataset))
or more compactly:
multiGplot(defarray("LRM", "LRY"), "--time-series
--with-lines", \
deflist(dataset))
Of course the string array and the list can also be predefined and
named, instead of this anonymous usage.
The function returns a string with the produced gnuplot plot source for
further use, if needed.
I've also thought if instead of (or in addition to) the list input a
bundle should be used which could also hold matrices and such things.
Comments on this and other aspects are welcome. Bugs are surely there.
cheers,
sven
6:30 p.m.
On Fri, 25 May 2018, Andreas Zervas wrote:
Dear all,
I am sending this email because lately I have observed changes in
the way gretl interacts with octave. I am attaching a simple program
for illustrative purposes. I think that such a program should work
in any case.
I think it works in linux with no problems (based on earlier
experience with more detailed programs to produce graphs).
The issue is that I was able to run this program under windows 7
with earlier gretl and octave versions, in particular it works under
gretl version up to 2016d and old octave versions - works with
octave 3.4.3. However, under win 7 it does not work with late octave
versions - does not work with octave 4.2.1.
In any case, I cannot make it work under win 10, no matter which
version of gretl (I have used 1.9.92, 2016d, 2017d) and octave
(3.4.3, 4.2.1) I use.
Could you take a look on this issue?
I can confirm that your script works on Linux using current gretl and
Octave 4.4.0. I'll try testing on Windows when I get a chance.
Allin Cottrell
8:02 p.m.
On Fri, 25 May 2018, Allin Cottrell wrote:
On Fri, 25 May 2018, Andreas Zervas wrote:
[about a gretl
script thet calls Octave to make a PDF plot]
>
> In any case, I cannot make it work under win 10, no matter which version of
> gretl (I have used 1.9.92, 2016d, 2017d) and octave (3.4.3, 4.2.1) I use.
>
> Could you take a look on this issue?
I can confirm that your script works on Linux using current gretl and Octave
4.4.0. I'll try testing on Windows when I get a chance.
I just tried with current gretl and Octave 4.4.0 on Windows 10. the
script ran OK, except that it wrote the output file, plotx.pdf (which
was given without a path in the script) into my gretl "dotdir"
(temporary storage area) so it was not readily findable, and moreover
would be automatically deleted on exit from gretl.
That issue is fixable by specifying a path to the plot file in the
script, but I'd say it is a gretl bug. We should ensure we're in the
gretl "workdir" when calling a "foreign" program so the user knows
where this sort of output is going.
By the way, the path to Octave that I specified under gretl's
Preferences was
C:\Octave\Octave-4.4.0\bin\octave-cli.exe
(after accepting their default installation path).
Allin
2396
days inactive
2407
days old
4 comments
3 participants
participants (3)
-
Allin Cottrell -
Andreas Zervas -
Sven Schreiber