OpenEuphoria

Re: Phix+EuGTK

• Posted by petelomax Jan 29, 2021
• 1521 views

Yes, it's probably better as well as easier and faster to just test it first:

--if Box(bx) then       -- best, if/when you can 
--if struct(bx) then    -- or Widget or whatever (see note below) 
if is_struct(bx,0) then -- (equivalent to struct) 
--if is_struct(bx,Box) then -- same as Box() 
--if is_struct(bx,routine_id("Box")) then -- ditto 
    ?bx.handle 
end if 

However, I think you know this, there will (probably) come a day when it succeeds, because {"struct","Box",5,2} or close enough has been re-created for something else, and you or someone else will sorely wish it hadn't.
Actually, though, there is something I could and should do to improve upon that: implement a new builtin (and use it in is_struct(), and therefore in both struct() and Box() too) [DONE, ready for next release]

bool bRes = still_has_delete_routine(object x) 

I'd still recommend doing everything you can to nullify all delete()'d instance refs/vars, tho.

Update: Added still_has_delete_routine(), used it, and fixed the glitch in using struct() ready for the next release.
I also cleaned up the original comments and added a couple extra options for testing.

Should you hit that struct() bug, the fix is in pmain.e, DoSubScripts() line 6495:

--31/1/21: 
--      integer stype = symtab[N][S_ltype] 
        integer stype = symtab[N][S_vtype]