Re: printing two touching, side-by-side plots leaves space between them

[Ben Abbott]

On May 27, 2011, at 11:51 AM, Martin Senator wrote:
> On Tue, May 17, 2011 at 03:02:10PM -0400, Ben Abbott wrote:
>> On May 17, 2011, at 2:25 PM, Martin Senator wrote:
>>> On Thu, Apr 28, 2011 at 05:10:47PM -0400, Ben Abbott wrote:
>>>> On Apr 28, 2011, at 4:33 PM, Martin Senator wrote:
>>>> 
>>>>> To: address@hidden
>>>>> Cc: ms
>>>>> Subject: printing two touching, side-by-side plots leaves space between 
>>>>> them
>>>>> --------
>>>>> Bug report for Octave 3.2.2 configured for i486-pc-linux-gnu
>>>>> 
>>>>> Description:
>>>>> -----------
>>>>> 
>>>>> I created two side-by-side plots with a common y scale that touch
>>>>> each other (using a procedure that Ben Abbot showed me).
>>>>> The plots were touching on the screen.
>>>>> When I used the print command, the resulting plot had a horizontal
>>>>> space between the two parts.
>>>>> 
>>>>> Repeat-By:
>>>>> ---------
>>>>> Following is the contents of file test1.m:
>>>>> 
>>>>> # testing the procedure Ben Abbot proposed:
>>>>> # file test1.m;
>>>>> # going to run by $ cat test1.m |octave --silent 
>>>>> # just issued the (octave) commands 'diary on, echo on, more off'
>>>>> 
>>>>> diary on, echo on, more off
>>>>> # testing the procedure Ben Abbot proposed:
>>>>> # file test1.m;
>>>>> # going to run by $ cat test1.m |octave --silent 
>>>>> # just issued the (octave) commands 'diary on, echo on, more off'
>>>>> 
>>>>> figure(1)
>>>>> clf
>>>>> x=0:0.01:10;
>>>>> h1=axes();
>>>>> h2=axes();
>>>>> p1=p2=get(h1,"position");
>>>>> p1(3)= p2(3)= p1(3)/2;
>>>>> p2(1)=p2(1)+p1(3);
>>>>> axes(h1)
>>>>> plot(x,sin(x))
>>>>> xlabel("radians")
>>>>> ylabel("sin(x)")
>>>>> axes(h2)
>>>>> plot(x*180/pi,sin(x))
>>>>> xlabel("degrees")
>>>>> set(h1,"position",p1);
>>>>> set(h2,"position",p2,"ytick",[]);
>>>>> outfile = "test1.pdf" ;
>>>>> print(outfile)
>>>>> pause(20)
>>>>> diary off
>>>> 
>>>> Using the developers sources, this problem/bug has been fixed. Plot is 
>>>> attached.
>>>> 
>>>> Ben
>>>> 
>>> 
>>> Ben,
>>> The original bug was on Octave 3.2.2.
>>> 
>>> I downloaded and built Octave 3.4.0,
>>> the current stable version.
>>> It also has the bug.
>>> 
>>> Is downloading and building the developer's source
>>> that doesn't have the bug the same order of complexity
>>> to an unsophisticated user like me?
>>> If so, what 'developer's version' do I look for?
>>> 
>>> Thanks,
>>> 
>>> Martin
>> 
>> I verified that the bug it not fixed in Octave 3.4.0.
>> 
>> However it has been fixed in the 3.4.x branch. You can download either the 
>> stable and default branch by following the instructions at the link below.
>> 
>>      http://www.gnu.org/software/octave/download.html
>> 
>> Be sure to scroll down to the "Development Sources" section. Once you have 
>> the sources you'll need to run "./autogen.sh" from the source directory and 
>> then do the same as for 3.4
>> 
>> Ben
> 
> Ben,
> I am having a problem with building the development sources.
> 
> I installed `mercurial' and ran the `hg clone ...' command,
> which ran to completion.  Then I changed to a newly formed
> subdirectory, $ cd octave ;
> then I ran $ ./autogen.sh ;
> it ran ./bootstrap, which reported that one of glibtoolize
> or libtoolize is required.
> I then installed gnulib and ran ./autogen.sh again.
> This time ./bootstrap required `git'.
> I installed git-core and ran ./autogen.sh again.
> This time the last message I received was
> ./bootstrap: getting gnulib files...
> Initialized empty Git repository in
> /home/ms/octave/oct-dev/octave/gnulib/.git/
> [on the next line just the cursor]
> and the process stalled.  Doing a 'ps afxw' command
> (in another window)
> I saw that ./autogen.sh had status S+ and was the parent of
> ./bootstrap, which also had status S+ and was the parent of
> git clone --depth 2 git://git.sv.gnu.org/gnulib gnulib,
> which also had status S+ and was the parent of two processes:
> an exact copy of itself (with status S+) and
> git index-pack --stdin -v --fixthin --keep=fresh-pack 20536 on senator2
> (20536 was the process id of its parent).
> 
> The first time it hung, the subdirectory gnulib/.git existed and, with
> its subdirectories, was empty (I am almost sure).
> The two subsequent times (I tried again twice) no subdirectory gnulib
> existed.
> While it was hung, I tried ping'ing git.sv.gnu.org, 
> and had success.
> 
> What should I do next?
> 
> Thanks,
> 
> Martin

As there are others who are more knowledgeable than I on the details of 
building Octave, I've cc'd the maintainers list.

Any chance you behind a firewall that is blocking port 9418?

If so you can try http cloning ...

        https://github.com/blog/92-http-cloning

The less than obvious syntax to get this to work (h/t to Jarno) is ...

        git clone http://git.savannah.gnu.org/r/gnulib.git

You can place the gnulib sources anywhere you like, and then point configure to 
it by setting the GNULIB_SRCDIR variable. For a bash shell, I set the variable 
prior to running configure.

        export GNULIB_SRCDIR="/Users/bpabbott/Development/git/gnulib"

If port blocking isn't the problem, then I'm in over my head.

Ben