[Info-vax] ODS-5 specifications, was: Re: Does OpenVMS Use Unicode?

[Stephen Hoffman]

On 2016-06-15 13:26:10 +0000, hb said:

> On 06/15/2016 02:38 PM, Stephen Hoffman wrote:
>> the linker will need changes if you allow UTF-8 symbols,
> 
> Yes and no. The linker takes a symbol as is and it simply matches them 
> as is to resolve undefined symbols. IF you want a map with the symbol 
> names printed in a readable fashion, then the linker has to change.

So... yes.  The linker will have to change.

If any sort of upcasing or downcasing eventually hits the linker to 
deal with that (related) mess, there are additional changes lurking for 
the linker, too.  But I digress.

On 2016-06-14 20:04:28 +0000, hb said:

> I agree - and the above example is more of a proof that it is too 
> complex than a proof of a "nice" feature. You named it "ODS-5 escaping" 
> but it's rather "DCL escaping" as ODS-5 does not see - and does not 
> need - any of this escaping. Have a look at a DUMP/HEADER of one of the 
> "escaped" filenames and you will see.

That the filenames are stored using laser-etched cuneiform is an 
implementation detail.   But between the ^ escaping and the ^UP^ and 
the pseudonames — and those wonderful DECC logical names misused as a 
configuration mechanism that can control or destabilize various apps 
entirely out-of-band — more than a few OpenVMS command procedures, as 
well as other interpreters and compiled apps have to be aware of the 
rules.  There's unfortunately not even a conversion or fallback API or 
related around either, AFAIK.


-- 
Pure Personal Opinion | HoffmanLabs LLC