gnuplot development

Ethan Merritt wrote:
> Unless there are objections, I propose to change the PATCHLEVEL
> and other embedded names in the 4.2 CVS tree to be 4.2 patchlevel 0,
> and prepare an official distribution source package this weekend.

No objection from me, but a little concern about us not having been able 
to reach Thomas Williams for the formal "blessing".  We would be going 
out a bit further on that limb we've manoeuvered ourselves onto 
regarding the License issue.

> 1) put the 4.2 release tarball on SourceForge
> 2) add pointers and update links on the web site
> 3) let those propagate
> 4) wait a bit (2 days? a week?)

3 and 4 shouldn't amount to more than 1 day combined.  SourceForge has 
sufficiently efficient mirroring systems to allow it.  The file release 
by itself will already alert a certain number of subscribers directly by 
mail.  Others can wait until we reach step 5)

> 5) publish the release announcement    (Draft text follows)
>  
> I was not involved in the mechanics of previous releases,
> so maybe there is an well-trodden release path that I am unaware of?

Given we've only used those paths thrice in 5 years (since version 
3.7.2), i doubt any of them could be classified as well-trodden. ;-)

I would put the announcement on Freshmeat and as a News entry on our 
SourceForge.net pages.  Ethan should post the announcement to our Usenet 
newsgroup, and I assume Petr will forward it to the Octave crowd.


> Short list of fixes since the final release candidate 4.2-rc4
> -------------------------------------------------------------

I don't really think this should be in the announcement.  Or if it goes 
in, then so should the entire NEWS since version 4.0.

Hans-Bernhard Bröker wrote:
> Ethan Merritt wrote:
> 
>>Unless there are objections, I propose to change the PATCHLEVEL
>>and other embedded names in the 4.2 CVS tree to be 4.2 patchlevel 0,
>>and prepare an official distribution source package this weekend.
> 
> 
> No objection from me, but a little concern about us not having been able 
> to reach Thomas Williams for the formal "blessing".  We would be going 
> out a bit further on that limb we've manoeuvered ourselves onto 
> regarding the License issue.

Yes, and if one is able to reach him, it would be good to discuss some way of resolving the License issues...  I could try contacting the Colin Kelly at that Callwave company to see if he is the other half of the original copyright holders.  However, any identifiable info about Kelly that Thomas Williams could provide would help, e.g., a college he may have been attending or worked for at the time.


>>Short list of fixes since the final release candidate 4.2-rc4
>>-------------------------------------------------------------
> 
> 
> I don't really think this should be in the announcement.  Or if it goes 
> in, then so should the entire NEWS since version 4.0.

If this were to be tagged 4.2-rc5 (as well as 4.2) then perhaps.  But for a major release version number, not really necessary.

Dan

Here's an item I just noticed:

        Type `help` to access the on-line reference manual.
        The gnuplot FAQ is available from
                http://www.gnuplot.info/faq/

        Send comments and help requests to  <gnuplot-beta@...>
        Send bug reports and suggestions to <gnuplot-beta@...>

The two email addresses are the same, is that intended?  If so, maybe:

        Type `help` to access the on-line reference manual.
        The gnuplot FAQ is available from
                http://www.gnuplot.info/faq/

        Send comments, help requests, bug reports and suggestions to
                <gnuplot-beta@...>

Dan

On Fri, Mar 02, 2007 at 07:07:47PM -0600, Daniel J Sebald wrote:
> Here's an item I just noticed:
> 
> The two email addresses are the same, is that intended?  If so, maybe:
> 
>         Type `help` to access the on-line reference manual.
>         The gnuplot FAQ is available from
>                 http://www.gnuplot.info/faq/
> 
>         Send comments, help requests, bug reports and suggestions to
>                 <gnuplot-beta@...>
> 
> Dan

and I just noticed the FAQ says:
   Please contribute your suggestions
   with respect to the file faq.tex available from
   http://cvs.sourceforge.net/viewcvs.py/gnuplot/faq/.

which address does not seem to work.

Also the bug reporting advice is inconsistent - i.e., no mention of the
bug tracker in the FAQ or gnuplot.doc 'Seeking Assistance' though Bugs
does mention uploading a report.  What is the preferred channel?

Gnuplot.doc has a 2004 copyright date.  Is revising the copyright
date there, in Copyright, and elsewhere part of the release plan?

I would add OpenVMS as an OS for which mouse support is incomplete.

- Lucas Hart

On Tuesday 27 February 2007 06:49, Kuscsik Zoltan wrote:
> automake: configure.in: required file `../config.guess' not found
> automake: configure.in: required file `../config.sub' not found

Those files are part of the automake package, not the gnuplot
distribution. What version of automake and you using?
Are you sure it (automake) was installed completely?

-- 
Ethan A Merritt

Kuscsik Zoltan wrote:
> but ./prepare says:
[...]

> automake: configure.in <http://configure.in>: required file 
> `../config.guess' not found
> automake: configure.in <http://configure.in>: required file 
> `../config.sub' not found

Most likely your automake/autoconf tools are outdated.  gnuplot does not 
require config.sub and config.guess.

I had automake1.4 installed. After updating to 1.9 config.guess and
config.sub
were no longer required. Thank you for your help.

Zoltan Kuscsik

On 2/28/07, Ethan Merritt <merritt@...> wrote:
>
> On Tuesday 27 February 2007 06:49, Kuscsik Zoltan wrote:
> > automake: configure.in: required file `../config.guess' not found
> > automake: configure.in: required file `../config.sub' not found
>
> Those files are part of the automake package, not the gnuplot
> distribution. What version of automake and you using?
> Are you sure it (automake) was installed completely?
>
> --
> Ethan A Merritt
>

On Monday 26 February 2007 19:20, m sutton wrote:
> I was looking at margins.dem and noticed the 'set multiplot title' command.  
> Then I looked in the doc and it was not listed.  Is this just an oversight? 

Sees that it was overlooked.
Thanks.


-- 
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742

On Sunday 25 February 2007 03:36, Harald Harders wrote:
> Dear all,
> 
> maybe, this mailing list is the wrong address for this question, but I 
> don't know better, at the moment.
> 
> Today, somebody requested a link to the file ps_fontfile_doc.pdf (Using 
> TeX Fonts in the Gnuplot Postscript Terminal) in the WWW which is part of 
> the gnuplot distribution. Unfortunately, it is not accessable via the WWW 
> pages below http://gnuplot.sourceforge.net/. I would appreciate if the 
> document could be made downloadable either on the "Documentation" or the 
> "Tutorials, learning, help" page.
> 
> Could you please let me know what you think about that?

As you say, it is part of the distribution.
Do we really need to post separate copies of files that everyone should
already have?   I'm not objecting, but...


Anyhow, the web pages could use a thorough updating.
If you want to submit a revised page or pages, go right ahead.
The page http://gnuplot.sourceforge.net/documentation.html
in particular seems very incomplete.


-- 
Ethan A Merritt

> Today, somebody requested a link to the file ps_fontfile_doc.pdf (Using 
> TeX Fonts in the Gnuplot Postscript Terminal) in the WWW which is part of 
> the gnuplot distribution. Unfortunately, it is not accessable via the WWW 
> pages below http://gnuplot.sourceforge.net/. I would appreciate if the 
> document could be made downloadable either on the "Documentation" or the 
> "Tutorials, learning, help" page.
> 
> Could you please let me know what you think about that?

Good idea. I've found that file on my disk, installed by SUSE's yast from 
their rpm, but not by gnuplot's cvs "make install".

I think it could be added into "documentation.html" somewhere around
"Official gnuplot online documentation" ... e.g. "Official ... concerning 
postscript". Then, all those files from psdoc/ could be available there.

If you update that .html file and enclose all pdf/ps files in a zip file, 
which I could just unzip on sf, then please contribute.

---
PM

Petr Mikulik wrote:
>>> It seems to me the current largest show-stopper for gnuplot is its
>>> official web page, http://www.gnuplot.info. It is not mirroring
>>> gnuplot.sf.net for several months.
>>> Further, in January 2006, there was thread in this mailing list:
>>> Subject: Re: ftp://ftp.gnuplot.info/pub/gnuplot/testing
> 
> 
> Thanks for the update! http://www.gnuplot.info/ is now up-to-date.

Ok, forget parts of my previous message!  I've recognized that the links
contained in http://www.gnuplot.info are pointing to gnuplot.sf.net.

funnily the pages in gnuplot.info are not the same as in http://www.gnuplot.info!

Also there seems to be one picture missing:
http://www.gnuplot.info/grid.jpeg is not being found.

Hardy

Petr Mikulik wrote:
>>> It seems to me the current largest show-stopper for gnuplot is its
>>> official web page, http://www.gnuplot.info. It is not mirroring
>>> gnuplot.sf.net for several months.
>>> Further, in January 2006, there was thread in this mailing list:
>>> Subject: Re: ftp://ftp.gnuplot.info/pub/gnuplot/testing
> 
> 
> Thanks for the update! http://www.gnuplot.info/ is now up-to-date.

Unfortunately they are not in sync!  Check 
http://gnuplot.sourceforge.net/development/binaries/ and 
http://gnuplot.info/development/binaries/

Because I'm a little pedantic (;-)) I'm telling you that this is the 
case since mid of september 2006.  There is an entry in the mail list 
dated 24.09.2006 which states this.

Hardy

> the plot displays fine, but if I change the view interactively, the  
> color scale changes (to be less useful).  The plot is still accurate,  
> but the color scale changes so that
> the whole plot is basically the same color.

thanks for the bug report, I've fixed it.

---
PM

Mojca Miklavec wrote:
> windows, I found out that the web reference to Help Workshop mentioned
> in the makefile isn't working any more (I guess that the domain has
> changed). 

I'm afraid the same would apply to any change we make to this 
information, rather sooner than later.  Microsoft has made a sport out 
of completely rearranging their websites on a regular basis, meaning no 
deep link into it is likely to remain available for any useful amount of 
time.  The only stable location specifier would be "somewhere on the 
net, probably at Microsoft".

On 2/24/07, Hans-Bernhard Br=F6ker wrote:
> Mojca Miklavec wrote:
> > windows, I found out that the web reference to Help Workshop mentioned
> > in the makefile isn't working any more (I guess that the domain has
> > changed).
>
> I'm afraid the same would apply to any change we make to this
> information, rather sooner than later.  Microsoft has made a sport out
> of completely rearranging their websites on a regular basis, meaning no
> deep link into it is likely to remain available for any useful amount of
> time.  The only stable location specifier would be "somewhere on the
> net, probably at Microsoft".

I don't visit Microsoft pages very often, but hardcoding the link
indeed doesn't seem too reliable.

But it makes no sense to keep broken/non-existing links in comments
either. I would mention that one should probably look for
"hcwsetup.exe" on Microsoft pages (in addition to what is already
written). It's not evident from makefile.mgw whether that is a
Microsoft product or some third-party tool. I wouldn't be surprised if
some tool on some X page would promote that name, but contain a virus
instead.

But it's only a minor (unimportant) change after all, no need to loose
time on it.

Mojca

Ethan A Merritt wrote:
> On Saturday 24 February 2007 02:04, Petr Mikulik wrote:
> 
>>Bugs item #1655001, was opened at 2007-02-08 10:18
>>You can respond by visiting: 
>>https://sourceforge.net/tracker/?func=detail&atid=102055&aid=1655001&group_id=2055
>>
>>Ethan, now this works differently in gnuplot 4.2rc4 and 4.3cvs:
>>
>>Is it intended? Could you please apply your patch for 4.3 also for 4.2?
> 
> 
> I do not want to make any more code changes for 4.2.
> Last minute configuration fixes or documentation tweaks only.
> Let's get it out the door!

Agreed.

> This patch ("splot_map_border") is one of a dozen or so that I
> have been collecting against 4.2.
> Each is self-contained and already in cvs, but was developed
> after the 4.2 code freeze. I give a complete list below.
> 
> My thought was to bundle these with fixes for any serious bugs
> reported for the distributed 4.2 code, and make them available
> several months after 4.2 goes out as "patch level 1".

If you want, but will patch levels get the consideration that a new release does to justify the effort?  (I'm not sure how software bundles are maintained.)  Otherwise, use your judgement of what to move into the release.  If you want help verifying behavior of any of these, just make a  short comment of what command or demo to look at for each:

>  1) rgbvar2D              Allow 3rd col of color info in 2D plots
>  2) image_clip            Implements image clipping in post pdf wxt gd
>  3) pdf_fonts             error trap and recovery for missing font info
>  4) offset_ambiguity      reduce the problem of trailing comma after 2D offset
>  5) 3dimpulses            splot with impulses lt rgb variable
>  6) zero_width_sample     if sample box is 0 width, don't draw boundary either
>  7) svg_pathlength        avoid ridiculously long lines in svg output
>  8) clipping_failure      fixes bug #1649887 that triggered a segfault when
>                           line segment endpoint went negative in bitmap.c
>  9) splot_map_border      Force border in 'set view map' to be either front or back
> 10) gd_grid               Draw dotted grid as alternating current_color/transparent
> 11) label_def_pt          Provide a default point type for labels
> 12) int_error_reset       Call back into datafile.c from int_error() to reset context
> 13) strcol                Allow strcol() as shorthand for stringcolumn()

Dan