I've now committed what I have to svn. Fixing the output format to %.20g
in pltclgen.tcl results in example 29 at least working. The output is
different to the C version though. To me this looks a little like rounding
errors, but I'm not sure. Perhaps fixing this up properly as you suggest
will remove the remaining issues.
Andrew
On Mon, Aug 24, 2009 at 08:31:57AM +0200, Arjen Markus wrote:
> Hi Andrew,
>
> I realised later that we can use Tcl_SetObjResult() instead of
> Tcl_SetResult() without migrating to the full Tcl_Obj interface style
> yet. That will allow us to return double precision floating point
> results without the string conversion. But fixing that issue you
> mentioned should take priority. What is it?
>
> Regards,
>
> Arjen
>
> On 2009-08-21 22:11, Andrew Ross wrote:
> > Hi Arjen,
> >
> > This helps - I no longer get a crash. There is still a further issue
> > somewhere. I'll see if I can track it down.
> >
> > Andrew
> >
> > On Fri, Aug 21, 2009 at 04:20:29PM +0200, Arjen Markus wrote:
> >> Hi Andrew,
> >>
> >> I looked at the source code for the bindings and I realised that
> >> currently we do not use tcl_precision or something similar to _return_
> >> values from PLplot routines. Instead we simply use %f to format the
> >> returned number. Up to now that was satisfactory, but the time functions
> >> return data with high precision.
> >>
> >> As a temporary workaround, could you try with a format like %.20f -
> >> like:
> >>
> >> Line 283, 284 in pltclgen.tcl:
> >> "PLFLT& {
> >> puts $GENFILE " sprintf( buf, \"%.20f\", $argname($i) );"
> >>
> >> (I seem to have added all manner of comments about it - never got around
> >> to actually implement them. It was not pressing then. I will be glad
> >> to take care of it myself, once I can comfortably access my computer
> >> at home again)
> >>
> >> Regards,
> >>
> >> Arjen
> >>
> >> On 2009-08-19 13:41, Andrew Ross wrote:
> >>> Arjen,
> >>>
> >>> No problem. I've updated example 29, but because of rounding errors in tcl
> >>> it is crashing. I've disabled the call to plot4 for now.
> >>>
> >>> I think that xmin and xmax end up being very similar (differences of < 1
> >>> second
> >>> over ~1000 years. Tcl precision issues seem to make them identical which
> >>> causes
> >>> a zero x range for the plot and hence a crash.
> >>>
> >>> Are there any ways round this? I tried setting tcl_precision, but it made
> >>> no
> >>> difference.
> >>>
> >>> Andrew
> >>>
> >>> On Wed, Aug 19, 2009 at 09:09:38AM +0200, Arjen Markus wrote:
> >>>> Hi Andrew,
> >>>>
> >>>> thanks for taking care of the Fortran and Tcl bindings.
> >>>> (I would have done it myself, but my time at the keyboard
> >>>> at home is severely limited at the moment.)
> >>>>
> >>>> Regards,
> >>>>
> >>>> Arjen
> >>>>
> >>>> On 2009-08-18 17:51, Andrew Ross wrote:
> >>>>>> On Tue, Aug 18, 2009 at 11:05:31AM +0200, Werner Smekal wrote:
> >>>>>>> Ok,
> >>>>>>>
> >>>>>>> I submitted now the last D example and the D bindings are therefore
> >>>>>>> finished. Actually not really finished, since example 20 delivers
> >>>>>>> slightly different postscript results than the C example. Although if
> >>>>>>>
> >>>>>>> I compare the output (xwin, wxwidgets, psc) visually I can't spot any
> >>>>>>>
> >>>>>>> difference. I don't think this is a showstopper, but will have a look
> >>>>>>>
> >>>>>>> later on, I don't consider that as too serious.
> >>>>>>>
> >>>>>>> In addition I ran the plplot tests on Mac OS X with the following
> >>>>>>> results:
> >>>>>>>
> >>>>>>> Test timeout computed to be: 1500
> >>>>>>> c++
> >>>>>>> Missing examples :
> >>>>>>> Differing postscript output : 29
> >>>>>>> Missing stdout :
> >>>>>>> Differing stdout :
> >>>>>> Werner,
> >>>>>>
> >>>>>> This one I find odd. C++ results should be completely identical to the
> >>>>>> C
> >>>>>> ones since there is no worrying about cross-language support or any of
> >>>>>> that. I get identical results for all examples with C and C++ using
> >>>>>> latest
> >>>>>> svn.
> >>>>>>
> >>>>>>> java
> >>>>>>> Missing examples : 20 21 22 23 24 25 26 27 28 29 30 31
> >>>>>>> Differing postscript output : 19
> >>>>>>> Missing stdout :
> >>>>>>> Differing stdout : 19
> >>>>>>> python
> >>>>>>> Missing examples :
> >>>>>>> Differing postscript output :
> >>>>>>> Missing stdout :
> >>>>>>> Differing stdout : 23
> >>>>>>> tcl
> >>>>>>> Missing examples :
> >>>>>>> Differing postscript output : 03 16 19 21 24 26 29
> >>>>>>> Missing stdout :
> >>>>>>> Differing stdout : 21
> >>>>>>> d
> >>>>>>> Missing examples :
> >>>>>>> Differing postscript output : 20
> >>>>>>> Missing stdout :
> >>>>>>> Differing stdout :
> >>>>>>> -- Process completed
> >>>>>>> ***Failed
> >>>>>>>
> >>>>>>> 88% tests passed, 2 tests failed out of 17
> >>>>>>>
> >>>>>>> The following tests FAILED:
> >>>>>>> 3 - examples_java (Failed)
> >>>>>>> 17 - examples_compare (Failed)
> >>>>>>> Errors while running CTest
> >>>>>>>
> >>>>>>> The only bad error is the segmentation fault of the Java example 19.
> >>>>>>> Otherwise everything configures and compiles fine on Mac OS X.
> >>>>>>>
> >>>>>>> Next I'll improve the Lua samples, then pdf, then wxWidgets device.
> >>>>>
> >>>>> I have now implemented all but the plslabelfunc call in f77 / f95
> >>>>> bindings
> >>>>> and updated the examples accordingly. I've left the plslabelfunc call
> >>>>> to
> >>>>> last as it is a bit messy. Octave is also up to date (except for
> >>>>> example 19
> >>>>> - not yet worked out how to do function callbacks for mapform /
> >>>>> labelfunc.)
> >>>>>
> >>>>> tcl clearly still needs work, as do ada and ocaml. Otherwise, we're
> >>>>> beginning
> >>>>> to get all the languages and examples back into shape again ready for
> >>>>> the
> >>>>> release.
> >>>>>
> >>>>> Andrew
> >>>>>
> >>>>> ------------------------------------------------------------------------------
> >>>>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008
> >>>>> 30-Day
> >>>>> trial. Simplify your report design, integration and deployment - and
> >>>>> focus on
> >>>>> what you do best, core application coding. Discover what's new with
> >>>>> Crystal Reports now. http://p.sf.net/sfu/bobj-july
> >>>>> _______________________________________________
> >>>>> Plplot-devel mailing list
> >>>>> Plplot-devel@lists.sourceforge.net
> >>>>> https://lists.sourceforge.net/lists/listinfo/plplot-devel
> >>>>>
> >>>> ------------------------------------------------------------------------------
> >>>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008
> >>>> 30-Day
> >>>> trial. Simplify your report design, integration and deployment - and
> >>>> focus on
> >>>> what you do best, core application coding. Discover what's new with
> >>>> Crystal Reports now. http://p.sf.net/sfu/bobj-july
> >>>> _______________________________________________
> >>>> Plplot-devel mailing list
> >>>> Plplot-devel@lists.sourceforge.net
> >>>> https://lists.sourceforge.net/lists/listinfo/plplot-devel
> >>>>
> >> ------------------------------------------------------------------------------
> >> Let Crystal Reports handle the reporting - Free Crystal Reports 2008
> >> 30-Day
> >> trial. Simplify your report design, integration and deployment - and focus
> >> on
> >> what you do best, core application coding. Discover what's new with
> >> Crystal Reports now. http://p.sf.net/sfu/bobj-july
> >> _______________________________________________
> >> Plplot-devel mailing list
> >> Plplot-devel@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/plplot-devel
> >>
> >
>
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now. http://p.sf.net/sfu/bobj-july
> _______________________________________________
> Plplot-devel mailing list
> Plplot-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/plplot-devel
>
------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now. http://p.sf.net/sfu/bobj-july
_______________________________________________
Plplot-devel mailing list
Plplot-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/plplot-devel
