Re: Last call for issues affecting release of gnuplot 5.0

[Ethan M. <merritt@u.washington.edu>]

On Monday, 29 December 2014 09:53:27 PM Karl Ratzsch wrote:
> The "update" command
> 
> http://sourceforge.net/p/gnuplot/bugs/1385/
> 
> might still need a change that is not backward-compatible, although
> i´d wager nobody would ever notice ;-).

Urk.  I let that one slip.  Thanks for pointing it out.

The version of "update" in both 5.0 and 5.1 simply doesn't work,
for the reasons given in the bug tracker.

Back in September I reverted the equivalent broken code for 4.6.6
to the previous long-standing version used through 4.6.5.
The idea was to fix it properly for version 5 but that never happened.

I'll think about this overnight, but probably it's too late
to do anything other than revert the code in gnuplot 5.0 also.
That will make "update" will behave exactly as it did in gnuplot 4.6,
buggy or not.  Better to have backwards compatibility with quirks
than to have neither backwards compatibility nor working code.

> I´d propose to make it always overwrite the ".old" backup file
> (which is the (unintended) behaviour on linux now, but not e.g.
> windows).

Unfortunately the behaviour on linux now is that it cannot even find
the old file reliably, because the existfile() routine knows nothing
about search paths.  So it always writes a new file in the current
directory, regardless of whether that was wanted.

	Ethan

> 
> 
> (And while at it:)
> fit.c contains an stale variant for the case when the parameter file
> does not exist and no previous fit was done. Should the file with
> all parameters marked FIXED be written or not? I´d vote for the
> former. It seems rather arbitrary to deny it.
> ==fit.c:1521==
> #if 1
> Eex2("'update' requires a prior 'fit' since the parameter file %s
> does not exist yet.", ofilename);
> #else
> fprintf(stderr, "'update' without a prior 'fit' and without a
> previous parameter file:\n");
> fprintf(stderr, " all variables will be marked '# FIXED'!\n");
> #endif
> =========
> 
>   Karl
> 
> 
> On 29.12.2014 19:27, sfeam wrote:
> > Hi all,
> > 
> > I plan to package up gnuplot 5.0 and place the tarball on SourceForge
> > for release on New Year's Day.  If you know of any remaining issues 
> > that should be either fixed or mentioned in the Release Notes, please
> > post them here on the mailing list.
> > 
> > The current text of the Release Notes is here:
> > 
> >   http://gnuplot.sourceforge.net/ReleaseNotes_5_0.html
> > 
> > 		Ethan
> > 
> > 
> > 
> > ------------------------------------------------------------------------------
> > Dive into the World of Parallel Programming! The Go Parallel Website,
> > sponsored by Intel and developed in partnership with Slashdot Media, is your
> > hub for all things parallel software development, from weekly thought
> > leadership blogs to news, videos, case studies, tutorials and more. Take a
> > look and join the conversation now. http://goparallel.sourceforge.net
> > _______________________________________________
> > gnuplot-beta mailing list
> > gnu...@li...
> > Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-beta
> > 
> > 
> 
> 

-- 
mail:   Biomolecular Structure Center,  K-428 Health Sciences Bldg
        MS 357742,   University of Washington, Seattle 98195-7742