gnuplot-bugs — Place to submit bug reports on gnuplot

[Gnuplot-bugs] plot iteration is not accepted after assignment in gnuplot 5.0.5

[ivana r. <iva...@mf...>]

Hi all,

after the gnuplot update from v5.0.0 to v5.0.5, the plot iteration is
not recognize after an assigment:

plot 1./x, for [i=1:3] i*x  #ok in both versions
plot 1./x, a=3, for [i=1:3] i*x  #fails in the newer one
#with error: "undefined variable: for."
plot 1./x, a=3, 1/0 not, for [i=1:3] i*x  #ok in both versions
#but linestyles are shifted.

is there any covering patch or so?

thanks

Iva

Re: [Gnuplot-bugs] undocumented features of gnuplot call-argument parser

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

Am 18.12.2016 um 18:29 schrieb ivana richterova:

> reading the gnuplot manual I'd guess that the call arguments would be
> passed as are but they are numerically expanded and split before being
> transferred to strings ARG0..ARG9 if they are not strings themselves.

I wonder what part of the manual you gathered that guess from.

And no, the call arguments are not specially parsed at all.  That 
parsing you see is normal gnuplot command parsing, which expects numbers 
and numeric expressions by default.  If you wanted to pass strings, 
you're supposed to enclose them in strings.

[Gnuplot-bugs] undocumented features of gnuplot call-argument parser

[ivana r. <iva...@mf...>]

Hi all,

reading the gnuplot manual I'd guess that the call arguments would be
passed as are but they are numerically expanded and split before being
transferred to strings ARG0..ARG9 if they are not strings themselves.
(This not occures for ``gnuplot -c <script> ARGS''.)

with:
# cat show_arguments.gplot
print "ARGn=\"value\" #strlen(value)"
print "ARGC=", ARGC
do for [n=0:ARGC] {
        print what="ARG".n, "=\"", val = value(what), "\" #", strlen(val)
}
---------------------------------------

I get:
# (1)
# gnuplot -d -e 'call "show_arguments.gplot" ar1 2ar'
ARGn="value" #strlen(value)
ARGC=3
ARG0="show_arguments.gplot" #20
ARG1="ar1" #3
ARG2="2" #1
ARG3="ar" #2
---------------------------------------

i.e., the same as for
# gnuplot -d -e 'call "show_arguments.gplot" ar1 2 ar'
---------------------------------------

# (2)
# gnuplot -d -e 'call "show_arguments.gplot" et1 2et 4et'

call "show_arguments.gplot" et1 2et 4et
                                  ^
line 0: expecting exponent
---------------------------------------

also, call command accept just one more argument than it  can:

# gnuplot -d -e 'call "show_arguments.gplot" 1 2 3 4 5 6 7 8 9 10'
ARGn="value" #strlen(value)
ARGC=10
ARG0="show_arguments.gplot" #20
ARG1="1" #1
ARG2="2" #1
ARG3="3" #1
ARG4="4" #1
ARG5="5" #1
ARG6="6" #1
ARG7="7" #1
ARG8="8" #1
ARG9="9" #1
ARG10="NaN" #line 11: internal error : strlen of non-STRING argument
---------------------------------------

# gnuplot -d -e 'call "show_arguments.gplot" 1 2 3 4 5 6 7 8 9 10 11'

call "show_arguments.gplot" 1 2 3 4 5 6 7 8 9 10 11
                                                   ^
"show_arguments.gplot", line 0: too many arguments for 'call <file>'
---------------------------------------

(tested on
Linux deimos 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt9-3 (2015-04-23)
x86_64 GNU/Linux
gnuplot Version 5.0 patchlevel 5    last modified 2016-10-02
)

sincerely

Iva

[Gnuplot-bugs] colorsequence bugs

[ivana r. <iva...@mf...>]

Hi all,

on
#Linux deimos 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt9-3 (2015-04-23)
x86_64 GNU/Linux
#gnuplot Version 5.0 patchlevel 5    last modified 2016-10-02

I've noticed two things I guess not being intended:

(1)
# sh colorsequence
gives
#sh colornames
instead. (No way to get actual colorseq as far as I know.)

(2) save command does not keep the colorsequence setting (even as a
comment). IMO this violates the save command manual definition.

Try e.g.,
#test=`mktemp ./test.XXXXXXX`
## ploted in classic colors
#gnuplot -d -e 'se colors classic; p x; save "'$test'"'
## ploted in default colours
#gnuplot -d -p $test
## ploted in classic colors
#gnuplot -d -p -e 'se colors classic' $test

I didn't find these bugs reported here before so do myself.

Sincerely

Iva

[Gnuplot-bugs] How to disable shell calls

[ivana r. <iva...@mf...>]

Hi all,

I'd like to open this theme since the last topic post I found is from
2008 and unsolved as far as I know.

I guess the restrict mode should be able (as an option and at least to
be configured).

In restricted mode, all system calls (!, ``, system(), <) should be
disable and only the current directory (including subdirectories)
should be accessible for reading and writing files.

Thanks

Iva

[Gnuplot-bugs] Tikz terminal key opaque not working correctly

[Gilbert F. <gi...@er...>]

Hello,
using the tikz terminal with today's gnuplot (csv) boxes created with
"set key box opaque" are white filled in my pdfs generated using
pdflatex, but not on printer output (tested on two devices).

Replacing

\gpfill{color=gpbgfillcolor}
(1.380,1.165)--(3.584,1.165)--(3.584,3.167)--(1.380,3.167)--cycle;

by

\draw [fill=gpbgfillcolor]
(1.380,1.165)--(3.584,1.165)--(3.584,3.167)--(1.380,3.167)--cycle;

fixes this issue.

best regards
Gilbert Forkel

Re: [Gnuplot-bugs] gnuplot 5.1 patchlevel 0 fails to compile with wxwidgets

[Alan F. <al...@co...>]

On 10/16/2014 3:32 AM, Hans-Bernhard Bröker wrote:
> Am 16.10.2014 um 01:20 schrieb Alan Feuerbacher:
>> However, without the --disable-wxwidgets switch make fails with:
>>
>> #######
>> In file included from /usr/include/gtk-2.0/gdk/gdkscreen.h:32:0,
>>                     from /usr/include/gtk-2.0/gdk/gdkapplaunchcontext.h:31,
>>                     from /usr/include/gtk-2.0/gdk/gdk.h:32,
>>                     from wxterminal/wxt_gui.h:183,
>>                     from wxterminal/wxt_gui.cpp:96:
>> /usr/include/gtk-2.0/gdk/gdktypes.h:114:39: error: conflicting
>> declaration ‘typedef struct _GdkDrawable GdkWindow’
>>     typedef struct _GdkDrawable           GdkWindow;
>>                                           ^
>> In file included from /usr/include/wx-3.0/wx/wxprec.h:12:0,
>>                     from wxterminal/wxt_gui.h:75,
>>                     from wxterminal/wxt_gui.cpp:96:
>> /usr/include/wx-3.0/wx/defs.h:3412:31: note: previous declaration as
>> ‘typedef struct _GdkWindow GdkWindow’
>>         typedef struct _GdkWindow GdkWindow;
>>                                   ^
>
> That's a conflict between two libraries' header files about the
> definition of a type.  Which means those two libraries are incompatible
> with each other: that version of gtk cannot be combined with that
> version of wx.

I see. How can I know which versions are compatible with each other?

I read the installation documentation that's part of the tar files for 
each, and saw nothing about prerequisites or dependencies.

Alan

Re: [Gnuplot-bugs] gnuplot 5.1 patchlevel 0 fails to compile with wxwidgets

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

Am 16.10.2014 um 01:20 schrieb Alan Feuerbacher:
> However, without the --disable-wxwidgets switch make fails with:
>
> #######
> In file included from /usr/include/gtk-2.0/gdk/gdkscreen.h:32:0,
>                    from /usr/include/gtk-2.0/gdk/gdkapplaunchcontext.h:31,
>                    from /usr/include/gtk-2.0/gdk/gdk.h:32,
>                    from wxterminal/wxt_gui.h:183,
>                    from wxterminal/wxt_gui.cpp:96:
> /usr/include/gtk-2.0/gdk/gdktypes.h:114:39: error: conflicting
> declaration ‘typedef struct _GdkDrawable GdkWindow’
>    typedef struct _GdkDrawable           GdkWindow;
>                                          ^
> In file included from /usr/include/wx-3.0/wx/wxprec.h:12:0,
>                    from wxterminal/wxt_gui.h:75,
>                    from wxterminal/wxt_gui.cpp:96:
> /usr/include/wx-3.0/wx/defs.h:3412:31: note: previous declaration as
> ‘typedef struct _GdkWindow GdkWindow’
>        typedef struct _GdkWindow GdkWindow;
>                                  ^

That's a conflict between two libraries' header files about the 
definition of a type.  Which means those two libraries are incompatible 
with each other: that version of gtk cannot be combined with that 
version of wx.

[Gnuplot-bugs] gnuplot 5.1 patchlevel 0 fails to compile with wxwidgets

[Alan F. <al...@co...>]

Hopefully this list is appropriate for this issue. Apologies if not.

I recently installed wxidgets-3.0.1 and gnuplot-5.1 to my 
LinuxFromScratch system. No apparent issues installing wxwidgets.

For installing gnuplot I followed instructions from:

http://www.gnuplot.info/
http://sourceforge.net/projects/gnuplot/files/gnuplot/
http://www.gnuplot.info/development/index.html

I did the following to get and compile the source code:

export 
CVSROOT=:pserver:ano...@gn...:/cvsroot/gnuplot
cvs login
cvs -z3 checkout gnuplot

ln -sfv qt5.conf /etc/xdg/qtchooser/default.conf
export QT_SELECT=qt5
ln -sfv /opt/qt5/lib/qt5/bin /opt/qt5/bin

./prepare
./configure --prefix=/usr --with-tutorial --disable-wxwidgets
make

This compiled successfully and executed the demo programs without issues.

However, without the --disable-wxwidgets switch make fails with:

#######
In file included from /usr/include/gtk-2.0/gdk/gdkscreen.h:32:0,
                  from /usr/include/gtk-2.0/gdk/gdkapplaunchcontext.h:31,
                  from /usr/include/gtk-2.0/gdk/gdk.h:32,
                  from wxterminal/wxt_gui.h:183,
                  from wxterminal/wxt_gui.cpp:96:
/usr/include/gtk-2.0/gdk/gdktypes.h:114:39: error: conflicting 
declaration ‘typedef struct _GdkDrawable GdkWindow’
  typedef struct _GdkDrawable           GdkWindow;
                                        ^
In file included from /usr/include/wx-3.0/wx/wxprec.h:12:0,
                  from wxterminal/wxt_gui.h:75,
                  from wxterminal/wxt_gui.cpp:96:
/usr/include/wx-3.0/wx/defs.h:3412:31: note: previous declaration as 
‘typedef struct _GdkWindow GdkWindow’
      typedef struct _GdkWindow GdkWindow;
                                ^
Makefile:877: recipe for target 'wxterminal/wxt_gui.o' failed
make[3]: *** [wxterminal/wxt_gui.o] Error 1
make[3]: Leaving directory '/sources/gnuplot-4.6.5/src'
Makefile:936: recipe for target 'all-recursive' failed
make[2]: *** [all-recursive] Error 1
make[2]: Leaving directory '/sources/gnuplot-4.6.5/src'
Makefile:395: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/sources/gnuplot-4.6.5'
Makefile:331: recipe for target 'all' failed
make: *** [all] Error 2
#########

I have no clue what to do about the above. Is there a bug in the 
compilation software for wxwidgets or gnuplot? I've not found anything 
on Google.

Thanks,
Alan

[Gnuplot-bugs] make installer fails on Mingw build

[Tatsuro M. <tma...@ya...>]

Hello

I have tried to make a installer for gnuplot for windows on MinGW bplatform (gcc-4.8.2, MinGW-w64 32bit) using cvs source (changelog  2014-03-20).  The make all and make install worded correctly.  Howerver, make installer failed.  The error messages are :

********************************************************************
PATCHLEVEL`-win32-setup/" \
            ../../win/gnuplot.iss > ./dist/gnuplot.iss
(cd ./dist; "/c/PROGRA~1/INNOSE~1/iscc/" gnuplot.iss)
Inno Setup 5 Command-Line Compiler
Copyright (C) 1997-2012 Jordan Russell. All rights reserved.
Portions Copyright (C) 2000-2012 Martijn Laan
Inno Setup Preprocessor
Copyright (C) 2001-2004 Alex Yackimoff. All rights reserved.

Compiler engine version: Inno Setup 5.5.4 (a)

[ISPP] Preprocessing.
[ISPP] Preprocessed.

Error on line 1 in d:\usr\Tatsu\mingw32work\gnuplot\gnuplotcvs\gnuplot\config\mingw\dist\gnuplot.iss: Text is not inside a section.
Compile aborted.
make: *** [installer] Error 2

********************************************************************

Top 13 lines of gnuplot.iss
**********************************************
;
; $Id: gnuplot.iss,v 1.7 2014/03/15 04:24:34 markisch Exp $
;
; GNUPLOT - gnuplot.iss
;
;[
; Copyright 2011,2012   Bastian Märkisch
;
; Permission to use, copy, and distribute this software and its
; documentation for any purpose with or without fee is hereby granted,
; provided that the above copyright notice appear in all copies and
; that both that copyright notice and this permission notice appear
; in supporting documentation.
**********************************************************

What is the error message "Text is not inside a section." ?

It will grateful for me if some suggestions will be given.

Regards

Tatsuro

[Gnuplot-bugs] Too many columns in using specification' error.\

[<mw...@us...>]

I was recently trying to make use of the ytic() using capability with a boxxyerrorbars plot. I get a 'Too many columns in using specification' error. Is this a bug?

Here is a sample case:

# test case for too many columns bug for boxxyerrorbars

xspc = 0.045
yspc = .48
plot '-' using 1:2:($1-xspc):($1+xspc):($2-yspc):($2+yspc):3:ytic(4) w boxxyerrorbars fs solid fc pal
.1 1 0 lab1
.2 1 1 lab1
.3 1 7 lab1
.4 1 2 lab1
.5 1 3 lab1
.6 1 4 lab1

.1 2 2 lab2
.2 2 3 lab2
.3 2 5 lab2
.4 2 0 lab2
.5 2 1 lab2
.6 2 2 lab2
e



Any ideas?

Mike Sutton

[Gnuplot-bugs] unexpected difference between plot and splot of binary files with unstructured data

[cédric o. <ced...@gm...>]

Hello,

After having asked the question on the gnuplot list, I think it may be
a possible bug so I post my question to this list.

I have found a interesting comment on the web explaining how to plot
unstructured mesh written in a binary file in gnuplot. I knew to do
that in ascii but I have been looking for the binary possibility for a
while.

The command is for instance the following

plot 'file.dat' binary record=(15,-1) format='%float' u 1:2:3 w l palette

I wrote the binary file file.dat using octave (fwrite(file,vector)).

The 15 is because I plot cells of 5 points of two coordinates + the
value at the point (so 5x3).

cell=[ptA ptB ptC ptD ptA]
vector=[ptAx ptAy ptAval ptBx ptBy ptBval ptCx ptCy ptCval  ptDx ptDy
ptDval  ptAx ptAy ptAval ...];

The plot command gives the results I would expect with an ascii file.
However, the splot command does not because some cells of the
unstructured mesh are now connected although they should not.

In addition, this behaviour for splot is obtained with gnuplot 4.6 patchlevel 0.

Crossing lines are also observed for the splot AND plot commands on my
older version 4.2 patchlevel 2. Therefore, it is seems a correction
has been done between the two versions but only for the plot command.

Can someone tell mu if it is a bug or I misuse gnuplot?

Best regards,

Cédric

[Gnuplot-bugs] first plot wrong with 'set tics front' and 'plot ... with boxes'

[Thomas S. <t.s...@fz...>]

gnuplot 4.6p3 linux

when plotting 'with boxes' with 'set tics front' set,
in the first plot the boxes start at the lower canvas
border. in the next (re)plot the boxes start correctly
at the lower border of the plot area.

gnuplot 4.7 doesn't have this problem.

sample script:

set tics front
set xrange [0:10]
plot x with boxes
pause -1 "boxes start at lower canvas border"
replot
pause -1 "boxes start at lower plot border"

[Gnuplot-bugs] mouse zoom and bind keys not working on Mac OS X

[Stig S. <sti...@gm...>]

Hi,

I recently installed gnuplot Version 4.6 patchlevel 3 on a Mac OS X Version
10.8.4. I also install XQuartz 2.7.4 to be able to use x11.

Unfortunately I'm no longer able to use the mouse or the bind keys with
X11. When I type 'show mouse' I get the info that 'mouse is on', but I
can't zoom in on a plot, or e.g., toggle the grid by pressing 'g' in the
drivers window. I've seen a few others having the same problem, e.g. at
https://groups.google.com/forum/#!topic/comp.graphics.apps.gnuplot/wDO75fNkojw
But I haven' been able to find an answer or a solution.

Is this a known problem? Anything that can be done about it?

Kind Regards,
-Stig Syndergaard

[Gnuplot-bugs] Clustered histogram doesn't render data file with one row

[Jason G. <jjg...@nc...>]

I believe I have found a bug where data from a one row data file doesn't
get rendered when multiple plots are used. Providing data and plot file
below. If you add a dummy line to the one line data file it plots fine. If
you only plot the one line data file it also is find. I want to cluster
based on f1 and f2.

data file 1 - f1.data
"Title" "Percent 1" "Percent 2"
Type1 87.896832 12.103168

data file 2 - f2.data
"Title" "Percent 1" "Percent 2"
Type2 73.676931 26.323069
Type3 73.983282 26.016718
Type1 26.836433

plotfile:
set title "Percentages"
set ylabel "Pecentage of Data"
set yrange [0:100]
set terminal postscript color
set output "test.ps"
set style data histogram
set style histogram rowstacked gap 1
set style fill solid border -1

set boxwidth .5
set key autotitle columnheader
set xtics rotate by -45

plot newhistogram "1" lt 1, "f1.data" u 2:xtic(1) title 'P1', '' using 3
title 'P2', \
newhistogram "2" lt 1, "f2.data" u 2:xtic(1) notitle, '' using 3 notitle

[Gnuplot-bugs] Possible xtics bug when using boxplot with separation and variable x position

[Adrian M. <adi...@ya...>]

Hi,

I'm trying to plot my data with boxplots (zero separation between them) and variable x position. The boxplots are ok, but the xtics are not, they are all overlapped under the first boxplot. I would have expected them to be each under its own boxplot.

Here's some code (and an attached data file):

set timefmt "%H:%M:%S"
set ydata time
set xtics auto
set style data boxplot
set style boxplot separation 0
plot "flat" using 1:2:(1):1

to make things more obvious, try with a small, nonzero separation:

set style boxplot separation 0.1
replot


It looks like the tics aren't taking into consideration the boxplot's x position.


Any ideas on how to fix this?


Sincerely,

Adrian

[Gnuplot-bugs] tiny bug when using set xdata time

[Javier G. <j.a...@gm...>]

Hi,

I have found out that when creating a plot with time format in the x-axis,
using comands

set timefmt "%s"
set xdata time

there is a difference in the use of the column specifier.

There is no problem by using the full form.

plot "timedata1" using ($1):($2), "timedata2" using ($1):($3) axes x1y2

works perfectly,

however when using as

plot "timedata1" using 1:2, "timedata2" using 1:3 axes x1y2

the time data is not properly interpreted.

I noticed that when the timedata1 and timedata2 files are the same then
there is no problem at all,

cheers,
  Javier

[Gnuplot-bugs] gnuplot Homepage typo?

[Jeroen De V. <fl...@gm...>]

Hi,

I noticed that the homepage lists the release of gnuplot 4.6.0 on date
08.03.2011 instead of 08.03.2012.
Probably a typo ;-)

with kind regards.

Jeroen

[Gnuplot-bugs] Bug: SVG terminal, enhanced text mode and backslashes

[Tilman S. <til...@gm...>]

Heyho everyone,

before you complain: I tried to post to your newsgroup, but I was not
successful. However, I do not have the time to dig into the newsgroup
system for one post, therefore I post to this mailing list.

I think I found a bug:

Using the svg terminal and enhanced text mode, I am not able to produce
a single backslash in the output:

plot [...] title '\tT' produces "tT",
title '\\tT' produces "\\tT",
title "\tT" produces "T",
title "\\tT" produces "tT",
title "\\\tT" produces "T",
title "\\\\tT" produces "\\tT".

However, I need the text "\tT" in my svg and I need enhanced mode to use
a smaller font size than the default size. It works with enhanced mode
off, but then I cannot change the title's font size. I checked the text
output with Inkscape.

I am using Gnuplot 4.4.4.

Bye,
Tilman Schröder

-- 
Secure communication?
GPG: 0xE387875E at keys.gnupg.net

[Gnuplot-bugs] Bug Report: executing a couple of commands in command-line

[akira a. <aki...@gm...>]

Dear gnuplot,

I found a bug in gnuplot 4.4 patchlevel 4.

The bug involves an error when I try to do the followings:
1. create a file of seven-character name containing a function of
five-character name defined with a tricky combination of brackets.
2. pass the following commands via command-line option "-e":
(i) load the file
(ii) set variables
(iii) save settings

> echo "f2345(x,a)=sin(-1.0+((x)))" > 1234567
> gnuplot -e 'load"1234567"; set size 1.0,1.0; set origin 0,0; save"-"' > /dev/null

load"1234567"; set size 1.0,1.0; set origin 0,0; save"-"
                         ^
line 0: invalid expression

Note that the file name and the function name do not matter as far as
seven-character file name
and five-character function name (or even four-character function name).
The commands are apparently correct in their syntax.

A spooky behavior of the bug is that the error only occurs with the
above sequence of commands as far as I checked.
In other word, the error does NOT occur if any of followings is
slightly changed.
(i) length of file name/function name
(ii) combination of "set" commands
(iii) command expression including what spacing matters
See the attached files in detail.

The bug is obviously present, although the tests that I try to figure
out the bug
do not necessarily cover a systematic check (and my example may not be minimal).

Not to mention, one can easily avoid the error by proper spacing
around brackets for instance
while the bug remains unfixed.


Thanks in advance,


Akira Akaishi

Re: [Gnuplot-bugs] [patch/gnuplot.1] Minor fixes

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

On 17.11.2011 22:50, Daniel Leidert wrote:

> - By default, "-" chars are interpreted as hyphens (U+2010) by groff,
> not as minus signs (U+002D). Since options to programs use minus signs
> (U+002D), this means for example in UTF-8 locales that one cannot cut
> and paste options.

Let's not completely forget the possibility that roff implementations 
other than groff might be used to process this document.

Re: [Gnuplot-bugs] [patch/configure.in] Check for lua.pc *and* lua5.1.pc

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

On 17.11.2011 22:34, Daniel Leidert wrote:
> Hi,
>
> On Debian and Ubuntu systems (and probably more) there is no pkg-config
> file lua.pc but there is a lua5.1.pc. I've attached a patch

Sorry to inform you, but none of your attachments made it to the list. 
I suggest you re-post them to the Bug Tracker on the web page.

[Gnuplot-bugs] [bug] makeinfo generated index.html: collision between top-file and "index" command description

[Daniel L. <dan...@gm...>]

Hi,

When running the makeinfo-command to create the HTML documentation in
docs/htmldocs the resulting index.html file is a collision between the
top node and the documentation of the 'index' command. Check it out.

Clicking the "Top" link on all pages (except) on index.html leads to
index.html. But there the link leads to a non-existing file. Also
index.html mixes the top "menu" structure and the description of the
index command.

Regards, Daniel

[Gnuplot-bugs] [patch/webify*.pl] Scripts still use obsoleted Perl 4 libraries

[Daniel L. <dan...@gm...>]

Hi,

The webify*.pl perl scripts use the (AFAIK) obsoleted and
near-to-be-removed (5.16) perl4 ctime.pl library. The attached patch
uses a core module of Perl 5 as replacement.

Regards, Daniel

[Gnuplot-bugs] [patch/gnuplot.1] Minor fixes

[Daniel L. <dan...@gm...>]

Hi,

The attached patch made against CVS fixes two issues:

- By default, "-" chars are interpreted as hyphens (U+2010) by groff,
not as minus signs (U+002D). Since options to programs use minus signs
(U+002D), this means for example in UTF-8 locales that one cannot cut
and paste options.

- A dot at the beginning of a line starts a comment. Thus the line is
not shown in your manual page. So in the patch the dot gets prepended by
a zero width space.

Regards, Daniel