Re: Pre-release testing version of gnuplot 6.0.5

[Bastian M. <bma...@we...>]

Builds fine on Windows with MSYS2/CLANG64 and MSYS2/CLANGARM64.

There's only a minor type issue in internal.c I noticed only know:  On 
Windows, intgr_t is a "long long int", not a "long int" (which would 
have only 32 bits). Hence the labs() call on line 1174 should really be 
llabs().  Is that fine for other platforms, too? Or should I add an 
ugly  #ifdef?

Best wishes,

   Bastian

Am 10.07.2026 um 05:56 schrieb Ethan A Merritt:
> I have placed a testing version of the source release
> package for gnuplot 6.0.4 on the "testing" section of the
> SourceForge download site.
>
>      https://sourceforge.net/projects/gnuplot/files/gnuplot/testing/
>
> My tentative plan is to announce the actual release in about
> two weeks (end of July).  That can be pushed back if you find
> any problems or want more time to test on other platforms.
>
> Brief release notes below, or see
>       gnuplot.sourceforge.net/ReleaseNotes_6_0_5.html
>
> %%%%%%%%%%    6.0.5 Release Notes July 2026   %%%%%%%%%%%%%%%%
>
> New features
>
> * Full mousing support for multiplots
>    - all panels report mouse coordinates
>    - all panels respond to pan/zoom/rotate mouse movement and hotkeys
>    - "replot" reexecutes multiplot commands starting from current state
>    - "remultiplot" restores the original state before reexecuting multiplot
>    - interactive mousing remains active after ^C or "reset"
>
> * Unicode escape sequences \U+xxxx are accepted in more contexts
>    - escape sequences are expanded in "noenhanced" strings
>    - escape sequences are expanded during substring evaluation
>    - escape sequences are expanded in string functions strstrt() and strlen()
>
> * New built-in function prod [<var> = <start> : <end>] <expression>
>    - implements the sequential product operation
>               prod [i=a:b] f(i)   =   f(a) * f(a+1) * ... * f(b)
>
> * Data files in EUC-JP encoding can be processed from other environments
>    - If running in a EUC-JP locale, use "set encoding locale"
>    - If running in a different locale, use "set encoding eucjp"
>    - In either case, this affects only input data.  It does not change
>      the encoding (usually UTF-8) written to plots on output.
>
> * New coloring option for contourfill
>    - Arbitrary discrete color ranges can be define without using a special palette
>          set contourfill defined [zmin1:zmax1] color1, [zmin2:zmax2] color2, ...
>
> * Allow "string"[i] as shorthand for "string"[i:i]
>
> Changed behaviour
>
> * Trailing whitespace on the command line is now ignored on input.
>    This means that the line-continuation character '\' is effective even
>    if it is followed by an unintentional blank.
>
> * Octal escape sequences must conform to the documented form \ooo
>    - E.g. \7 or \0106 or \456 or \0 are now rejected
>
> * "plot with labels" now accepts the "noenhanced" keyword
>
> Bug fixes
>
> * qt: Slow font initialization on Windows                       #Bug# 2316 2476
> * gd: Resolved long-standing font handling problems.            #Bug# 2819
>    - requires recent library versions (libgd 2.4.0 libraqm 0.10.3)
> * Better tic placement on logscale axis spanning two decades    #Bug# 2856
> * Always clear previous plot title before a new plot is started #Bug# 2865
> * Endpoint errors due to floating point precision               #Bug# 2768
> * Repaired axis range parsing in parametric mode broken by change in 6.0.4
> * Reworked logic for extending the command buffer incrementally on input.
>    Long input lines with backslash continuation could lead to corruption
>    of the command string if encountered early in program execution.
>
> %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
>
> cheers,
> 	Ethan
>
>
>
>
>
> _______________________________________________
> gnuplot-beta mailing list
> gnu...@li...
> Membership management via: https://lists.sourceforge.net/lists/listinfo/gnuplot-beta