10:45 p.m.
Here's a follow-up to
http://lists.wfu.edu/pipermail/gretl-devel/2017-December/008296.html
et seq.
Some package-writers (and maybe others) have written hansl code to
generate custom plots that are difficult or impossible to obtain
via the built-in "gnuplot" and "plot" commands.
Gretl can _display_ any such plot provided the generated gnuplot
script is valid, but offering GUI editing (e.g. via right-click on a
plot window) is trickier. We can offer the editing facility -- with
a reasonable chance that this doesn't reduce the plot to rubble --
only if gretl "understands what's going on" in the plot (that is, if
we're able to parse enough information out of the plot commands to
make sense of it).
There's no way that gretl is going to incorporate a fully-fledged
gnuplot parser but some of my recent changes make the parsing
somewhat more flexible, hence increasing the chances that a custom
plot can be handled in the GUI plot editor. So I'd like to invite
authors of custom plotting code to try this out: Is your custom plot
now (nominally) editable? (That is, does the "Edit" option appear on
right-click?) And if so, does it actually work or does making
changes destroy the plot?
For anyone interested in pursuing this, here are a few pointers.
1) Take a look at the gretl-generated gnuplot code for a variety of
plot types, to get a sense of the "house style" which is likely to
be supported.
2) If C doesn't scare you away, maybe look at some of the functions
in gui2/gpt_control.c, in particular parse_gp_line_line() and
parse_gp_set_line(). In such functions you see gretl trying to make
sense of core gnuplot commands -- that should give you a more
precise idea of what formulations we can handle (and by elimination,
what we're not ready to handle).
3) If your plot contains formulations that gretl cannot handle
natively, try protecting them by placing them within the following
two special comment lines in the gnuplot file:
# start literal lines
...
# end literal lines
That way gretl will know not to mess with them.
Allin
11:25 p.m.
Many thanks Allin and nice timing!
I am currently trying to implement a few gnuplot plots via gretl functions. If somebody
has an interest, here is a current question related to automated "jitter" plots
https://stackoverflow.com/q/47865637/3895476 and here a git-hub repository with my current
code https://github.com/FrederikSchaff/gretl_scripts . It is a bit messy yet, but I try to
push only things that work (so far).
Perhaps it would be nice to share private scripts via git-hub or similar, even if not yet
in a format for a function package.
Best
Frederik
-----Ursprüngliche Nachricht-----
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Allin Cottrell
Gesendet: Dienstag, 19. Dezember 2017 23:45
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: [Gretl-devel] more plotting news
Here's a follow-up to
http://lists.wfu.edu/pipermail/gretl-devel/2017-December/008296.html
et seq.
Some package-writers (and maybe others) have written hansl code to
generate custom plots that are difficult or impossible to obtain via the built-in
"gnuplot" and "plot" commands.
Gretl can _display_ any such plot provided the generated gnuplot script is
valid, but offering GUI editing (e.g. via right-click on a plot window) is trickier.
We can offer the editing facility -- with a reasonable chance that this doesn't
reduce the plot to rubble -- only if gretl "understands what's going on" in
the
plot (that is, if we're able to parse enough information out of the plot
commands to make sense of it).
There's no way that gretl is going to incorporate a fully-fledged gnuplot
parser but some of my recent changes make the parsing somewhat more
flexible, hence increasing the chances that a custom plot can be handled in
the GUI plot editor. So I'd like to invite authors of custom plotting code to try
this out: Is your custom plot now (nominally) editable? (That is, does the
"Edit" option appear on
right-click?) And if so, does it actually work or does making changes destroy
the plot?
For anyone interested in pursuing this, here are a few pointers.
1) Take a look at the gretl-generated gnuplot code for a variety of plot types,
to get a sense of the "house style" which is likely to be supported.
2) If C doesn't scare you away, maybe look at some of the functions in
gui2/gpt_control.c, in particular parse_gp_line_line() and
parse_gp_set_line(). In such functions you see gretl trying to make sense of
core gnuplot commands -- that should give you a more precise idea of what
formulations we can handle (and by elimination, what we're not ready to
handle).
3) If your plot contains formulations that gretl cannot handle natively, try
protecting them by placing them within the following two special comment
lines in the gnuplot file:
# start literal lines
...
# end literal lines
That way gretl will know not to mess with them.
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
6:10 p.m.
It would be nice if one could save a custom plot (gnuplot scoped in a function) as icon
via "NAME <- ". Here an example:
<hansl>
function void test (series a, series b)
gnuplot a b
end function
t1 <- test(aa,bb)
t2 <- gnuplot aa bb
</hansl>
In fact, the "<-" even changes the file, by adding (this becomes obvious when
one tries to manually save it to the icon view via the context menu from the opened
graph):
<>
set print "C:/Users/Frederik/AppData/Roaming/gretl/gretltmp.png.bounds"
print "pixel_bounds: ", GPVAL_TERM_XMIN, GPVAL_TERM_XMAX, GPVAL_TERM_YMIN,
GPVAL_TERM_YMAX
print "data_bounds: ", GPVAL_X_MIN, GPVAL_X_MAX, GPVAL_Y_MIN, GPVAL_Y_MAX
</>
perhaps a special return type "gnuplot" or such would work?
-----Ursprüngliche Nachricht-----
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Allin Cottrell
Gesendet: Dienstag, 19. Dezember 2017 23:45
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: [Gretl-devel] more plotting news
Here's a follow-up to
http://lists.wfu.edu/pipermail/gretl-devel/2017-December/008296.html
et seq.
Some package-writers (and maybe others) have written hansl code to
generate custom plots that are difficult or impossible to obtain via the built-in
"gnuplot" and "plot" commands.
Gretl can _display_ any such plot provided the generated gnuplot script is
valid, but offering GUI editing (e.g. via right-click on a plot window) is trickier.
We can offer the editing facility -- with a reasonable chance that this doesn't
reduce the plot to rubble -- only if gretl "understands what's going on" in
the
plot (that is, if we're able to parse enough information out of the plot
commands to make sense of it).
There's no way that gretl is going to incorporate a fully-fledged gnuplot
parser but some of my recent changes make the parsing somewhat more
flexible, hence increasing the chances that a custom plot can be handled in
the GUI plot editor. So I'd like to invite authors of custom plotting code to try
this out: Is your custom plot now (nominally) editable? (That is, does the
"Edit" option appear on
right-click?) And if so, does it actually work or does making changes destroy
the plot?
For anyone interested in pursuing this, here are a few pointers.
1) Take a look at the gretl-generated gnuplot code for a variety of plot types,
to get a sense of the "house style" which is likely to be supported.
2) If C doesn't scare you away, maybe look at some of the functions in
gui2/gpt_control.c, in particular parse_gp_line_line() and
parse_gp_set_line(). In such functions you see gretl trying to make sense of
core gnuplot commands -- that should give you a more precise idea of what
formulations we can handle (and by elimination, what we're not ready to
handle).
3) If your plot contains formulations that gretl cannot handle natively, try
protecting them by placing them within the following two special comment
lines in the gnuplot file:
# start literal lines
...
# end literal lines
That way gretl will know not to mess with them.
Allin
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
2558
days inactive
2560
days old
2 comments
2 participants
participants (2)
-
Allin Cottrell -
Schaff, Frederik