What are your most urgent features in GDL ? (missing or weak or slow)

[alaingdl]

I think it is time to discuss what is weak or missing in GDL now. On my side most of the functionalities are working fine & fast, but yes, basically I use basic operations on data cube + The Astro Lib (FITS I/O) + MPfit

Missing :

• shmmap and shmvar does not exist in GDL #860 SHMMAP and SHMVAR
• Keyword parameter SHADES not supported in call to: SHADE_SURF #1605 some keywords give bad outputs in SHADE_SURF
• Any chance to implement FILE keyword in loadct? #1389 missing FILE keyword in LOADCT

Weak :

• better communication on the good quality of current GDL (the 1.0.4 !)
• since IDL 8.5 we are slower in some matrix operations (eg INVERT())
• UFT8 (unicode) problem #1676 UFT-8 related (should be possible because PLplot & WxWidgets are suppose to handle)
• some operations are slow on M1 & M2 (why ? only compiler issue ?!)
• when compiled with -Wall, huge amounts of warnings ...
• graphical tests ? (no provocation, just to mention we have to time time and systematize that)
• automatic tests on the PS outputs (no provocation, just to mention we have to time time and systematize that)
• CHOLDC/CHOLSOL exact behavior (partially OK) + tests are missing CHOLDC/CHOLSOL tests are missing #1311

Bugs :

• SURFACE (and SHADE_SURF) do not play nicely with !p.multi #1521 SURFACE & SHADE_SURF + !p.multi
• several bugs in TIMESTAMP() #1519 TIMESTAMP (I have a draft somewhere, but few tricky bugs remain :sad:)
• problems with POLY_FIT & test_poly_fit #1413 bug(s) or detail in POLY_FIT & test_poly_fit
• TRIRED #230 TRIRED related family functions
• ... (I did not review all the issues ! Some may be closed)

To be discussed, corrected and continued ...

Alain

[slayoo]

I'd also prioritise bringing back execution of Python-GDL tests contributed by @olebole on CI (this is a regression since we moved from Travis to GHA), see: #1055

[GillesDuvert]

At the moment I'm experimenting save/restore of procedures. No need to emphasise the interest.

Shmmap should be next as it's easy.

[GillesDuvert]

If gdl becomes handy, then the gdlde and the jupyter interest should be revived. Hope we'll get new contributors.