gnuplot-info — For questions and discussion about gnuplot

Re: [Gnuplot-info] Fwd: Memory overload in real time plotting

[Ethan A M. <EAM...@gm...>]

> The memory growth is observed in the program not gnuplot. This is the
> code
> that is being called periodically every few seconds. extents_plot_handler
> = gnuplot_init(); is only called once at some previous point.
> 
> int i;
> 
> 	char * cmd_header = "plot \"-\" matrix with image\n";
> 	char * cmd = malloc(
        ^^^^^^^^^^^^^^^
Why malloc() rather than realloc()?

> 	
> 			sizeof(char)
> 			
> 					* ((4 * extents_map_size)
> 					
> 							+ (int) (extents_map_size / pixels_width) + 30));
> 	
> 	strcpy(cmd, "");
> 	cmd = concat(cmd, cmd_header);
	
       ^^^^^^^^^^^^^^^^^^^^^^
The concat() function is not standard C, so I don't know exactly
what it does. But at a guess, possibly not what you think it does.
Try using strcat() instead.  
That's assuming you know for sure that there will not be a buffer
overrun.  

	Ethan

	
>	for (i = 1; i <= extents_map_size; i++) {
> 	
> 		cmd = concat(cmd, addr_space_extents_array[i - 1]);
> 		if (i % pixels_width == 0) {
> 		
> 			cmd = concat(cmd, "\n");
> 		
> 		}
> 	
> 	}
> 	if (extents_map_size % pixels_width > 0) {
> 	
> 		for (i = extents_map_size;
> 		
> 				i
> 				
> 						< (extents_map_size + pixels_width
> 						
> 								- (extents_map_size % pixels_width)); i++) {
> 			
> 			cmd = concat(cmd, "0 ");
> 		
> 		}
> 	
> 	}
> 	if (extents_map_size % pixels_width > 0)
> 	
> 		cmd = concat(cmd, "\ne\ne");
> 	
> 	else
> 	
> 		cmd = concat(cmd, "e\ne");
> 	
> 	/*FILE *file;
> 	
> 	 char file_name[20];
> 	 sprintf(file_name, "temp_data%d.txt", temp_counter++);
> 	 file = fopen(file_name, "a+");
> 	 fprintf(file, "%s", cmd);
> 	 fclose(file);*/
> 	
> 	fprintf(stderr,"extents_plot_handler size:
> %d\n",sizeof(extents_plot_handler));
> 
> 	fprintf(stderr,"extents_plot_handler file size:
> %d\n",sizeof(extents_plot_handler->gnucmd));
> 
> 	fprintf(stderr,"extents_plot_handler active plots:
> %d\n",extents_plot_handler->nplots);
> 
> 	fprintf(stderr,"extents_plot_handler temporary files:
> %d\n",extents_plot_handler->ntmp);
> 
> 	fprintf(stderr,"extents_plot_handler temp filename table size:
> %d\n",sizeof(extents_plot_handler->tmp_filename_tbl));
> 
> 	gnuplot_cmd(extents_plot_handler, cmd);
> 	free(cmd);
> 
> --
> View this message in context:
> http://gnuplot.10905.n7.nabble.com/Memory-overload-in-real-time-plotting
> -tp17795p17798.html Sent from the Gnuplot - User mailing list archive at
> Nabble.com.
> 
> -------------------------------------------------------------------------
> ----- Android is increasing in popularity, but the open development
> platform that developers love is also attractive to malware creators.
> Download this white paper to learn more about secure code signing
> practices that can help keep Android apps secure.
> http://pubads.g.doubleclick.net/gampad/clk?id=65839951&iu=/4140/ostg.clkt
> rk _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via:
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info

Re: [Gnuplot-info] Fwd: Memory overload in real time plotting

[tryfonaration <far...@gm...>]

The memory growth is observed in the program not gnuplot. This is the code
that is being called periodically every few seconds. extents_plot_handler =
gnuplot_init(); is only called once at some previous point.

int i;

	char * cmd_header = "plot \"-\" matrix with image\n";
	char * cmd = malloc(
			sizeof(char)
					* ((4 * extents_map_size)
							+ (int) (extents_map_size / pixels_width) + 30));
	strcpy(cmd, "");
	cmd = concat(cmd, cmd_header);
	for (i = 1; i <= extents_map_size; i++) {
		cmd = concat(cmd, addr_space_extents_array[i - 1]);
		if (i % pixels_width == 0) {
			cmd = concat(cmd, "\n");
		}
	}
	if (extents_map_size % pixels_width > 0) {
		for (i = extents_map_size;
				i
						< (extents_map_size + pixels_width
								- (extents_map_size % pixels_width)); i++) {
			cmd = concat(cmd, "0 ");
		}
	}
	if (extents_map_size % pixels_width > 0)
		cmd = concat(cmd, "\ne\ne");
	else
		cmd = concat(cmd, "e\ne");
	/*FILE *file;
	 char file_name[20];
	 sprintf(file_name, "temp_data%d.txt", temp_counter++);
	 file = fopen(file_name, "a+");
	 fprintf(file, "%s", cmd);
	 fclose(file);*/
	fprintf(stderr,"extents_plot_handler size:
%d\n",sizeof(extents_plot_handler));
	fprintf(stderr,"extents_plot_handler file size:
%d\n",sizeof(extents_plot_handler->gnucmd));
	fprintf(stderr,"extents_plot_handler active plots:
%d\n",extents_plot_handler->nplots);
	fprintf(stderr,"extents_plot_handler temporary files:
%d\n",extents_plot_handler->ntmp);
	fprintf(stderr,"extents_plot_handler temp filename table size:
%d\n",sizeof(extents_plot_handler->tmp_filename_tbl));
	gnuplot_cmd(extents_plot_handler, cmd);
	free(cmd);



--
View this message in context: http://gnuplot.10905.n7.nabble.com/Memory-overload-in-real-time-plotting-tp17795p17798.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

Re: [Gnuplot-info] plot dstat output cvs file

[Theo <th...@gm...>]

your data looks a bit distorted (return/newlines), maybe it's because of 
the mail program.

If I correct this and apply all your settings and

plot 'Data.csv' u 1:4 every ::6

with gnuplot 4.6.3 I do not have problems.

Do you get error messages? What is actually plotted?
Maybe try first simpler data to find out what's going wrong.

Theo.

Can you plot e.g. this?

"blabla1"
"blabla2"
"blabla3"
"blabla4"
"blabla5"
"blabla6"
30-10 11:19:40,1.0,2.0,3.0
30-10 11:19:41,1.1,2.1,3.1
30-10 11:19:42,1.2,2.2,3.2
30-10 11:19:43,1.3,2.3,3.3
30-10 11:19:44,1.4,2.4,3.4

[Gnuplot-info] Fwd: Memory overload in real time plotting

[Ethan M. <eam...@gm...>]

Please clarify.  Is the memory growth happening in gnuplot or in the
program at the other end of the pipe?   If the latter, I don't see how we
could make a guess at the problem without seeing your code.

I just had a very quick look at the code in gnuplot_1-2.11 from the URL you
mentioned.  It looks like the library itself allocates new memory on every
call to gnuplot_init() and frees it in gnuplot_close().   Are your calls to
these routines correctly paired?


On Wed, Oct 30, 2013 at 9:20 AM, tryfonaration
<far...@gm...>wrote:

> Hello,
>
> I am using a simple C library from here http://ndevilla.free.fr/gnuplot/to
> plot real-time data with gnuplot. It basically opens a pipe to gnuplot and
> you can feed it with data and commands. I want to plot an image that
> changes
> periodically. I make the following command string "plot '-' matrix with
> image", then concatenate a big chunk of data and then pipe it to gnuplot.
> This is done every a few seconds. What happens is that my program occupies
> more and more memory every time a new command is issued and eventually the
> OS kills it.
> Any idea why this could be happening?
>
>

[Gnuplot-info] Memory overload in real time plotting

[tryfonaration <far...@gm...>]

Hello,

I am using a simple C library from here http://ndevilla.free.fr/gnuplot/ to
plot real-time data with gnuplot. It basically opens a pipe to gnuplot and
you can feed it with data and commands. I want to plot an image that changes
periodically. I make the following command string "plot '-' matrix with
image", then concatenate a big chunk of data and then pipe it to gnuplot.
This is done every a few seconds. What happens is that my program occupies
more and more memory every time a new command is issued and eventually the
OS kills it. 
Any idea why this could be happening?





--
View this message in context: http://gnuplot.10905.n7.nabble.com/Memory-overload-in-real-time-plotting-tp17795.html
Sent from the Gnuplot - User mailing list archive at Nabble.com.

[Gnuplot-info] plot dstat output cvs file

[Bart V. <Bar...@te...>]

I'm trying to plot an output file from dstat obtained with

   dstat -tv --output dstat_timed.csv

the data-file looks as follows:

-----------------------------------------------------------------------------
"Dstat 0.7.2 CSV output"
"Author:","Dag Wieers 
<da...@wi...>",,,,"URL:","http://dag.wieers.com/home-made/dstat/"
"Host:","maverick",,,,"User:","mc303"
"Cmdline:","dstat -tv --output dstat_timed.csv",,,,"Date:","30 Oct 2013 
11:19:41 CET"

"system","procs",,,"memory 
usage",,,,"paging",,"dsk/total",,"system",,"total cpu usage",,,,,
"time","run","blk","new","used","buff","cach","free","in","out","read","writ","int","csw","usr","sys","idl","wai","hiq","siq"
30-10 
11:19:41,0.000,0.0,1.284,729153536.0,416268288.0,1697247232.0,1353175040.0,0.0,0.0,230338.748,250793.406,1390.199,2597.766,12.913,5.164,74.172,7.655,0.005,0.092
30-10 
11:19:42,0.0,0.0,0.0,729186304.0,416268288.0,1697263616.0,1353125888.0,0.0,0.0,0.0,0.0,955.0,1634.0,3.535,4.040,92.424,0.0,0.0,0.0
30-10 
11:19:43,0.0,0.0,1.0,729186304.0,416268288.0,1697263616.0,1353125888.0,0.0,0.0,0.0,0.0,1104.0,2043.0,5.612,4.082,89.796,0.0,0.0,0.510
30-10 
11:19:44,0.0,0.0,1.0,729194496.0,416268288.0,1697263616.0,1353117696.0,0.0,0.0,0.0,0.0,1071.0,1841.0,6.186,3.093,90.722,0.0,0.0,0.0
30-10 
11:19:45,0.0,0.0,17.0,730214400.0,416268288.0,1697263616.0,1352097792.0,0.0,0.0,0.0,4096.0,1313.0,2408.0,7.143,3.571,89.286,0.0,0.0,0.0
-----------------------------------------------------------------------------

I'm trying to plot with the following script:

---------------------------------------------------
set datafile separator ","

set xlabel "Time"
set xdata time
set timefmt "%d-%m %H:%M:%S"
set format x "%H:%M"

set title "procs"
plot "dstat_timed.csv" u 1:2 t columnheader(2) w lp, \
      "" using 1:3 t columnheader(3) w lp, \
      "" using 1:4 t columnheader(4) w lp
---------------------------------------------------

but unfortunately, this only works if I uncomment the first 6 lines in 
the data-file.

I already tried plotting with

   plot "dstat_timed.csv" u 1:2 every ::6 t columnheader(2) w lp

but this didn't help.

How can I tell gnuplot to ignore the first 6 lines in the data-file?

Thanks for your help,
Bart

Re: [Gnuplot-info] deleting gridlines

[Elias A. <eli...@gm...>]

On 10/26/2013 09:14 PM, Hans-Bernhard Bröker wrote:
> On 26.10.2013 03:23, tglahn17 wrote:
>
>> I made a graph with gridlines using the "set grid" command.  Is there a way to delete particular gridlines?  Thanks,
>
> Grid lines are coupled to axis tics.  You can always replace the
> automatically generated sequence of tics with your own selection (see
> "help set xtics"); the grid lines follow suit.

In addition, if you want to decouple the grid from the tics, you could 
draw your own grid using ‘set arrow’.

	Elias

Re: [Gnuplot-info] unwanted xtics

[BBands <bb...@gm...>]

set xtics ("program1" 0, "program2" 1, "program3" 2) nomirror

    John

On Mon, Oct 28, 2013 at 6:35 AM, Mahmood Naderan <nt_...@ya...> wrote:
> Hi,
>
> I have set the following commands to have the bottom and left brder lines
>
>    unset border
>    unset xtics
>    unset ytics
>    set border 3
>
> Now when I use this xtic formatting (for rotated strings), like this:
>
>    set xtics ("program1" 0, "program2" 1, "program3" 2)
>    set xtics rotate by 45 right
>
> I see the ticks on the top of the figure. In short, the top horizontal line is removed but the tics exist.
> How can I remove them?

Re: [Gnuplot-info] unwanted xtics

[Theo <th...@gm...>]

type 'help xtics'
and read the explanations carefully...

I guess 'nomirror' is what you are looking for...
set xtics nomirror

[Gnuplot-info] unwanted xtics

[Mahmood N. <nt_...@ya...>]

Hi,

I have set the following commands to have the bottom and left brder lines

   unset border
   unset xtics
   unset ytics
   set border 3

Now when I use this xtic formatting (for rotated strings), like this:

   set xtics ("program1" 0, "program2" 1, "program3" 2)
   set xtics rotate by 45 right

I see the ticks on the top of the figure. In short, the top horizontal line is removed but the tics exist.
How can I remove them?


Regards,
Mahmood

Re: [Gnuplot-info] deleting gridlines

[Hans-Bernhard B. <HBB...@t-...>]

On 26.10.2013 03:23, tglahn17 wrote:

> I made a graph with gridlines using the "set grid" command.  Is there a way to delete particular gridlines?  Thanks,

Grid lines are coupled to axis tics.  You can always replace the 
automatically generated sequence of tics with your own selection (see 
"help set xtics"); the grid lines follow suit.

[Gnuplot-info] deleting gridlines

[tglahn17 <tgl...@ya...>]

Hi,
 
I made a graph with gridlines using the "set grid" command.  Is there a way to delete particular gridlines?  Thanks,

Re: [Gnuplot-info] missing data = empty string ?

[Ethan M. <eam...@gm...>]

You have found a bug.
The problem is specifically the row that has an empty field in the first
column.  All the other empty fields are correctly recognized but that one
is not.

I don't have an immediate work-around to offer but I'm sure it's fixable.


On Fri, Oct 25, 2013 at 1:52 AM, Theo <th...@gm...> wrote:

> I have to bother you again.
> The missing data seems not work with matrix data.
>
> If I take the following TAB-separated matrix
>
> 1       2       3       4
> 2       3       4       5
> 3       4       5       6
> 4       5       6       7
>
> plot 'Matrix.dat' matrix with image
>
> I get the expected plot.
>
> However, if e.g. I leave out the 4s in the matrix
> 1       2       3
> 2       3               5
> 3               5       6
>         5       6       7
>
> and additionally apply:
>
> set datafile missing ""
> set datafile separator "\t"
>
> I get the error messages:
> "Matrix does not represent a grid"
> or
> "warning: matrix contains missing or undefined values"
>
> So, it seems that if values in the first line are missing it's not
> working...?
> Any idea to plot the data nevertheless without inserting '0' or 'NaN'?
>
>
>
>
>
>
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most
> from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60135991&iu=/4140/ostg.clktrk
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via:
> https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

Re: [Gnuplot-info] missing data = empty string ?

[Theo <th...@gm...>]

I have to bother you again.
The missing data seems not work with matrix data.

If I take the following TAB-separated matrix

1	2	3	4
2	3	4	5
3	4	5	6
4	5	6	7

plot 'Matrix.dat' matrix with image

I get the expected plot.

However, if e.g. I leave out the 4s in the matrix
1	2	3	
2	3		5
3		5	6
	5	6	7

and additionally apply:	

set datafile missing ""
set datafile separator "\t"

I get the error messages:
"Matrix does not represent a grid"
or
"warning: matrix contains missing or undefined values"

So, it seems that if values in the first line are missing it's not 
working...?
Any idea to plot the data nevertheless without inserting '0' or 'NaN'?

Re: [Gnuplot-info] xtics in logscale

[Hans-Bernhard B. <HBB...@t-...>]

On 21.10.2013 16:41, Mahmood Naderan wrote:

> I found that this command
>
>      set xtics (1,10,10000)
>
> works and is better than
>    set xtics 1,10
>    set xtics add (10000) Regards,

> I don't know why but that works.

Then you must re-read "help set xtics" until you do.  For starters let 
me just point out the obvious:

	set xtics 1,10

is not the same as

	set xtics (1, 10)

Re: [Gnuplot-info] xtics in logscale

[Mahmood N. <nt_...@ya...>]

>And what _is_ the output?

The output contains all tics  1,10,100,1000,1E+4,1E+5

I found that this command   
    
    set xtics (1,10,10000) 

works and is better than  
  set xtics 1,10
  set xtics add (10000) Regards,


I don't know why but that works.


>set format x "10^{%T}"

Yes that is also works.

Thanks
--------------------------------------------
On Mon, 10/21/13, Theo <th...@gm...> wrote:

 Subject: Re: [Gnuplot-info] xtics in logscale
 To: gnu...@li...
 Date: Monday, October 21, 2013, 5:44 PM
 
 I am not sure whether I unterstand
 you correctly.
 You may want to try
 
 set format x "10^{%T}"
 
 It will display the tic labels as 10^4, 10^5 etc. which
 saves space.
 Assuming you are using wxt terminal you can have a nicer
 look if you type:
 
 set term wxt enhanced
 
 
 Am 21.10.2013 15:52, schrieb Mahmood Naderan:
 > Hi
 > I have been confused by setting a custom tic in a
 logscale mode. On x-axis I have
 >
 > 1, 10, 100, 1000, 1E+4, 1E+5
 >
 > Problem is that in small figures, the 'E' tics overlap
 each other (because they are long) and the figure gets
 messy. So I wrote
 >
 > set logscale x
 > set xtics 1,10
 > set xtics add (10000)
 >
 > So I expect not to see 100, 1000 and 1E+5. But the
 output is not I want. What is wrong then?
 >
 >
 > Regards,
 > Mahmood
 >
 ------------------------------------------------------------------------------
 > October Webinars: Code for Performance
 > Free Intel webinars can help you accelerate application
 performance.
 > Explore tips for MPI, OpenMP, advanced profiling, and
 more. Get the most from
 > the latest Intel processors and coprocessors. See
 abstracts and register >
 > http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
 > _______________________________________________
 > gnuplot-info mailing list
 > gnu...@li...
 > Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
 >
 
 
 
 ------------------------------------------------------------------------------
 October Webinars: Code for Performance
 Free Intel webinars can help you accelerate application
 performance.
 Explore tips for MPI, OpenMP, advanced profiling, and more.
 Get the most from 
 the latest Intel processors and coprocessors. See abstracts
 and register >
 http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
 _______________________________________________
 gnuplot-info mailing list
 gnu...@li...
 Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
 

Re: [Gnuplot-info] xtics in logscale

[Theo <th...@gm...>]

I am not sure whether I unterstand you correctly.
You may want to try

set format x "10^{%T}"

It will display the tic labels as 10^4, 10^5 etc. which saves space.
Assuming you are using wxt terminal you can have a nicer look if you type:

set term wxt enhanced


Am 21.10.2013 15:52, schrieb Mahmood Naderan:
> Hi
> I have been confused by setting a custom tic in a logscale mode. On x-axis I have
>
> 1, 10, 100, 1000, 1E+4, 1E+5
>
> Problem is that in small figures, the 'E' tics overlap each other (because they are long) and the figure gets messy. So I wrote
>
> set logscale x
> set xtics 1,10
> set xtics add (10000)
>
> So I expect not to see 100, 1000 and 1E+5. But the output is not I want. What is wrong then?
>
>
> Regards,
> Mahmood
> ------------------------------------------------------------------------------
> October Webinars: Code for Performance
> Free Intel webinars can help you accelerate application performance.
> Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from
> the latest Intel processors and coprocessors. See abstracts and register >
> http://pubads.g.doubleclick.net/gampad/clk?id=60135031&iu=/4140/ostg.clktrk
> _______________________________________________
> gnuplot-info mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-info
>

Re: [Gnuplot-info] xtics in logscale

[Hans-Bernhard B. <HBB...@t-...>]

On 21.10.2013 15:52, Mahmood Naderan wrote:

> I have been confused by setting a custom tic in a logscale mode. On x-axis I have
>
> 1, 10, 100, 1000, 1E+4, 1E+5
>
> Problem is that in small figures, the 'E' tics overlap each other (because they are long)

Hmm.  But they're actually _shorter_ than your alternative, so how do 
you expect that to be any better?

 > and the figure gets messy. So I wrote
>
> set logscale x
> set xtics 1,10
> set xtics add (10000)
>
> So I expect not to see 100, 1000 and 1E+5.

How did you figure the above would remove the tics at 100, 1000 and 1e5?

 > But the output is not I want.

And what _is_ the output?

 > What is wrong then?

For now: your expectation.

[Gnuplot-info] xtics in logscale

[Mahmood N. <nt_...@ya...>]

Hi
I have been confused by setting a custom tic in a logscale mode. On x-axis I have

1, 10, 100, 1000, 1E+4, 1E+5

Problem is that in small figures, the 'E' tics overlap each other (because they are long) and the figure gets messy. So I wrote

set logscale x
set xtics 1,10
set xtics add (10000)

So I expect not to see 100, 1000 and 1E+5. But the output is not I want. What is wrong then?

 
Regards,
Mahmood

Re: [Gnuplot-info] missing data = empty string ?

[Theo <th...@gm...>]

Sorry, for the confusing previous post


OK, it seems to be solved... also in version 4.6.3

If you additionally enter

set datafile missing ""     # empty string

Only if you use both

set datafile missing ""
set datafile separator '\t'
plot 'Data.dat' u 1:($2) with lines, '' u 1:($3) with lines, '' u 1:($4) 
with lines

then it will draws the data as interrupted lines, as expected.

Re: [Gnuplot-info] missing data = empty string ?

[Theo <th...@gm...>]

OK, it seems to be solved... also in version 4.6.3

If you additionally enter

set datafile missing ""     # empty string


So you have to use both
Hi Ethan,

if you take the following simple datafile

0	2.0	3.0	4.0
1	2.1	3.1	4.1
2		3.2	4.2
3	2.3		4.3
4	2.4		4.4
5	2.5	3.5	4.5
6	2.6	3.6	
7	2.7	3.7	
8	2.8	3.8	4.8
9	2.9	3.9	4.9

set datafile separator '\t'
plot 'Data.dat' u 1:($2) with lines, '' u 1:($3) with lines, '' u 1:($4) 
with lines

You should get three parallel lines where all lines should be 
interrupted. As you can see in the plot, they are not.
I am using gnuplot 4.6.3
A version 4.6.4 or 4.7 windows executable is not (yet) available on 
sourceforge. So, alternatively, I would have to figure out how I can 
compile 4.6.4 or 4.7 on my computer. Haven't done it before. This might 
take a while.
Theo.

then it draws the data as interrupted lines, as expected.

Re: [Gnuplot-info] missing data = empty string ?

[Theo <th...@gm...>]

Hi Ethan,

if you take the following simple datafile 'Data.dat'

0	2.0	3.0	4.0
1	2.1	3.1	4.1
2		3.2	4.2
3	2.3		4.3
4	2.4		4.4
5	2.5	3.5	4.5
6	2.6	3.6	
7	2.7	3.7	
8	2.8	3.8	4.8
9	2.9	3.9	4.9

set datafile separator '\t'
plot 'Data.dat' u 1:($2) with lines, '' u 1:($3) with lines, '' u 1:($4) 
with lines

You should get three parallel lines where all lines should be 
interrupted. As you can see in the plot (I try to attach the plot), they 
are not.
I am using gnuplot 4.6.3
A version 4.6.4 or 4.7 windows executable is not (yet) available on 
sourceforge. So, alternatively, I would have to figure out how I can 
compile 4.6.4 or 4.7 on my computer. Haven't done it before. This might 
take a while.
Theo.

Re: [Gnuplot-info] missing data = empty string ?

[Ethan A M. <eam...@gm...>]

On Sunday, 20 October 2013 09:41:50 PM Theo wrote:
> unfortunately not...
> 
> help says:
> set datafile separator {"<char>" | whitespace}
> 
> default is whitespace
> isn't that either \s or \t ?
> 
> Anyway,
> set datafile separator "\t"
> does not help...

That's odd. It should.

After saying "set datafile separator '\t'" an empty column should
nevertheless be recognized as a column.  So values in column 4
should never bleed into column 3 as you described originally.

As to whether an empty column is treated as zero or missing or 
bad, I'm afraid that depends on the gnuplot version.

Current gnuplot (4.6.4 and 4.7) treat empty columns as missing data.
Earlier versions treated it as "bad" if you tried to plot it but 
(I think) properly ignored it if the columns was not mentioned in 
the using spec.

	Ethan

Re: [Gnuplot-info] missing data = empty string ?

[Theo <th...@gm...>]

unfortunately not...

help says:
set datafile separator {"<char>" | whitespace}

default is whitespace
isn't that either \s or \t ?

Anyway,
set datafile separator "\t"
does not help...

Re: [Gnuplot-info] (no subject)

[Hans-Bernhard B. <HBB...@t-...>]

On 20.10.2013 17:14, Chiara wrote:
>
> I'm trying to plot the histogram produced by the command latency -g name_file of Xenomai, but I obtained this message:
> "warning: skipping unreadable file "latency - gr", no data in plot.

That's because there is, indeed, no such file.  A command line is not a 
file.  The output of a command can be captured to a file, but that does 
not make that command itself a file.