[Gnuplot-bugs] bugs/problems in gnuplot 3.8k (cvs)

[Harald K. <ko...@sc...>]

Hi,

I'm now using gnuplot 3.8.0j for quite a while (and older versions for ages)
mostly for data visualisation, data fits, automated measurement analysis etc.
yesterday I picked your latest CVS version. here are a few bugs/problems
which I've found (most of them exist in 3.8.0j too), and realizing that
4.0 seems to be not too far away, it's high time to report some
(for me obvious) problems/bugs:



1) pdf driver:  that's a great new driver -- I really love that addition!

the pdf rendering looks good, but the PDF file is missing some information/hint/...
that the (default) plot size is _landscape_.  acroread displays the the diagrams
just fine, but when trying to print them to postscript file/printer with it's 
default page setup "portrait",  I get a small landscape diagram on a portrait page
with lots of white space above and below the plot.
when I switch the acroread page setup to "landscape", printout is ok.

somehow, other pdf tools can inform acroread to set the page setup to "landscape",
so printing is ok without twiddling with page setup settings first.


2) readline(?), SIGTSTP, and gnuplot_x11 communication  (and libreadline?)

I'm building gnuplot on various Linux plattforms (and hpux/aix/solaris/osf1/*bsd)
using the following setup
	
	./configure  --without-linux-vga --without-gd --without-lisp-files --without-tutorial --with-readline=gnu

some tests quite a while ago with gnuplot 3.8h.0 and 3.8j.0
seemed to show that some/all of the following suspend/resume problems
showhow relate to the use of "--with-readline=gnu" (both
libreadline 4.2 and 4.3), because the minimal builtin readline support
is far from being sufficient when working a lot "online" with
gnuplot while trying to analyze new data sets and twiddling
with best displays/data filters/fits/..., so not using
--with-readline=gnu is not an option:(



a) when suspending gnuplot with CTRL-Z  or sending SIGTSTP (not SIGSTOP, 
that's working fine!) while no X11 plot window is open/active, 
gnuplot exits silently.  try:

	gnuplot
	CTRL-Z
	fg

now gnuplot exits without error (strace shows regular rundown).

	gnuplot
	plot sin(x)
	CTRL-Z
	fg
	replot

is no problem, but

	gnuplot
	plot x
	set term postscript
	CTRL-Z
	fg

again kills gnuplot (tried on various Linux versions with recent kernels).




b) if gnuplot with "active" x11 display (no other "set term ...") 
gets suspended with CTRL-Z and while being stopped, you move 
the mouse a bit over the gnuplot_x11 window, the gnuplot_x11 
(and X server) start to suck all CPU time until gnuplots
gets resumed (and sometimes (not always), one has to move 
the mouse across the gnuplot_x11 window againon)

strace on gnuplot_x11 shows the following sequence while looping:

	18:18:01.409522 select(4, [0 1 3], NULL, NULL, NULL) = 1 (in [1])
	18:18:01.413039 write(3, "\177\30\1\0", 4) = 4
	18:18:01.420691 ioctl(3, FIONREAD, [0]) = 0
	18:18:01.421109 write(1, "\0\0\0\0\t\1\0\0&\1\0\0\0\0\0\0\0\0\0\0P\376\5\10\320\356"..., 120) = -1 EAGAIN (Resource temporarily unavailable)


just an idea:
strace shows that gnuplot_x11 gets both SIGTSTP and SIGCONT,
maybe this can be used to disable/enable the mouse support ? 


c) another related problem is that after suspend/resume of gnuplot
with hanging X11 display (can also happen when gnuplot_x11 gets
redraw events because it was hidden/uncovered by other windows)
is that the gnuplot user terminal/command interface  hangs (no input possible)
after resume until the mouse pointer got moved over the gnuplot_x11
window. only after "touching" gnuplot_x11 unlocks the gnuplot
command interface.



3) I'm just starting to use gnuplot on M$ Win-XY for my first time.
obviously it's great too, but by default I'm hardly missing the 
possibility to use shell commands in plot "datafile" using 

	plot "< some command"

but by default PIPES config is not set for windows builds :-(

I've read your comments about PIPES in config/makefile.mgw
but for me (Windows novice) it's unclear if this is a problem
for _all_ windows versions (e.g. would WinXP be any better?).

any chance that you'll change that to PIPES=1 for windows 
by default, or to supply two win32 binaries with both settings ?

right now I still hope (dream?) that it'll be possible to use MSYS
to allow "real" shell commands in plot "< ..."


I'll be happy to give more details if necessary or test whatever 
patches/changes  you'd like being tested...



anyway, thanks for your great work and the wonderful tool gnuplot !!

Harald Koenig
-- 
"I hope to die                                      ___       _____
before I *have* to use Microsoft Word.",           0--,|    /OOOOOOO\
Donald E. Knuth, 02-Oct-2001 in Tuebingen.        <_/  /  /OOOOOOOOOOO\
                                                    \  \/OOOOOOOOOOOOOOO\
                                                      \ OOOOOOOOOOOOOOOOO|//
Harald Koenig                                          \/\/\/\/\/\/\/\/\/
science+computing ag                                    //  /     \\  \
ko...@sc...                            ^^^^^       ^^^^^