Re: [GLE-general] Problems with QGLE on windows 10

[Francois T. <fra...@gm...>]

A few belated comments --

1. About QGLE: I am on Debian (minimal netinst) and I could never run QGLE
successfully, it always complains about not finding a Ghostscript library.
This is a mismatch between ghostcript versions, QGLE looks for an older
version of ghostscript than the one installed on current Debian. I guess I
could fix this by downgrading my ghostscript libraries to older versions
but I never went through the trouble. I think of GLE as basically a
command-line application, no need for a gui.

2. It is important, though, to have fast visual feedback when writing a GLE
script. Any scriptable text editor (Vim, Emacs, Geany ...) will make it
easy to run GLE on the script file and then call a viewer to see the
results immediately. All of this on a single hotkey. I have been using vim
with a custom plugin, but I recently switched to Kakoune (which is amazing
and that I recommend to all vimmers!).

3. About GLE dying... No, it won't die, it is too good to die! I have never
seen such a combination of powerful, precise graphical results with ease of
scripting. What is true is that GLE should be much better known. In my
view, the fault lies with the web site. It is dated, many parts are wrong,
the opening claims even give wrong ideas about GLE (e.g., the writing on
the opening page and even in the guide gives the impression that one needs
Latex to write legends, which is untrue; etc.). Worst are the graph
galleries. Some plots are fantastic, but many others range from meh to
fugly. To sum up: in my view, revamping the web site is the most important
thing to do if the GLE community wants to gain traction. Last winter I
contacted Vince Labella about this, he has been quite helpful and I have
started to work on my side on the web pages. Basically all of the pages
should be rewritten, reorganized, etc. and most important all graph
galleries should be changed for the best. For starters, and most urgently,
the opening pages should be updated. We can keep a link to the old
galleries, and update/remove them one by one in a second, longer pass.
Eventually the GLE web site should look as good as matplotlib's.

4. I have been working hard on all of this, but have been unable to go
further during the last 4 months because of competing obligations with my
own research. I do hope we'll be able to update the web site later this
year, by december. Be ready for really good-looking pages and plots :-).


On Thu, Aug 15, 2019 at 9:57 AM Laurence Abbott via Glx-general <
glx...@li...> wrote:

> On Wed, 14 Aug 2019 at 19:46, Christian T. Steigies <ct...@de...>
> wrote:
>
>> On Tue, Aug 13, 2019 at 05:58:39PM +0200, Zbigniew Kisiel wrote:
>> > Hi All,
>> >
>> > I'm also an ASCII old timer over several different operating systems
>> and gle
>> > implementations.
>>
>> Interesting. My former advisor was using GLE on a DOS machine, and I think
>> later on windows. So half of the people on the list had to support their
>> boss with GLE? :-)
>>
>
> Ditto!
>
> We just like the way it creates nice, clean graphs and does what it's told!
>
> > I now use gle in two ways:
>> > 1/ my graphics programs automatically generate gle scripts of what they
>> > display
>> > 2/ diagrams are reedited from some old version (such as that from
>> automatic
>> > output)
>>
>> I started to use tikz to do this directly in LaTeX. This may be even more
>> powerful than GLE, except maybe for the data plotting part. For this I
>> still
>> prefer GLE over all other tools.
>>
>
> I'd not seen tikz before: I might find that handy for the odd thing but
> there's no way my boss would move to latex as well as a new plotting
> program!
>
>
>> I am not suggesting to drop QGLE, I am sure it is a useful tool. However
>> the
>> next Debian release (in two years?) may not support Qt4 apps anymore. So
>> either QGLE gets ported to Qt5, or Debian/bullseye will have GLE without
>> QGLE. Maybe it is not so hard, but I have never written a Qt app. These
>> suggestions seem to fix already half of the problems:
>>
>> https://www.ics.com/blog/porting-desktop-applications-qt-4-qt-5
>>
>> This seems to take a little more work:
>> https://stackoverflow.com/questions/26180311/qhttp-in-not-available-in-qt5
>
>
> I had a quick read around yesterday and (allegedly) porting from Qt4 to
> Qt5 has been designed to be fairly straightforward.
>
> I might try to have a fiddle over the next few days, although it would
> mean that I'd have to delve into the guts of autoconf, etc. to do it
> properly! I tried a quick test yesterday (on Debian) just changing
> qmake-qt4 to qmake-qt5 within the qgle directory and that did create me a
> makefile. Compilation then died pretty quickly. I suspect this was mostly
> due to missing qt5 -dev packages. I'm hoping that a suitably prepared
> configure script will shout at me clearly what ones are missing, rather
> than installing a load of random ones on the off chance!
>
> As an aside, I tested the qgle that was already built and it "mostly
> worked" with ghostscript 9.27: it worked fine for one gle file but
> immediately caused a seg fault with another one! I might also look into any
> recent API changes in ghostscript...
>
> Cheers,
>
> Laurence
> --
> Dr Laurence Abbott <lau...@yo...>
> Department of Chemistry,
> University of York,
> YO10 5DD, UK
> _______________________________________________
> Glx-general mailing list
> Glx...@li...
> https://lists.sourceforge.net/lists/listinfo/glx-general
>