very weird results when printing fstat results of files using a funcion

[brandy125]

I could not yet break down the source of error but it is somehow related to fstat and using it as a return value of a function.
Look this weired example:

openw,1,'erasemelater'
d=1.0
writeu,1,d
close,1

Now lets check the size of that file (the result should be 4 bytes):

bash$ gdl
  GDL - GNU Data Language, Version v1.1-42-gc19e3233

GDL> print,printsize('erasemelater')
% Compiled module: PRINTSIZE.
SIZE            LONG      =  -1232656200
 -1232656200
GDL> 
bash$ gdl
  GDL - GNU Data Language, Version v1.1-42-gc19e3233
GDL> print,printsize('erasemelater')
% Compiled module: PRINTSIZE.
SIZE            LONG      =  -1620301640
 -1620301640
GDL> 

Using the funcion printsize.pro:

function printsize,fname
    openr,dev,fname,/get_lun
    s=fstat(dev)
    size=s.size
    help,size
    return,size
end

In addition to that behaviour comes that if we put a stop inside the function then the value of size is correct.

Version used:

GDL> !gdl
{
    "RELEASE": "1.1-42-gc19e3233",
    "BUILD_DATE": "Feb  5 2025",
    "EPOCH": 1738724400,
    "GDL_USE_DSFMT": 1,
    "GDL_USE_WX": 1,
    "GDL_POSIX": 1
}
GDL> 

[jtappin]

Very odd! But it gets stranger.

If you add a free_lun, dev then the second (and subsequent) calls give zero as the size!

P.S. file_info is working correctly.

[GillesDuvert]

I made a change recently that seems to have adverse effects.... Thanks for the reports, patching is on its way!

[GillesDuvert]

#1982 solves it.