Re: Re: Cobol compiler date

---------------------------<snip>--------------------------------------

>
> > But IFOXxx was never part of OS/360.
> > It appeared with the first release of MVS.
>
> First release of SVS and VS1, I believe. And VM/370.
>
---------------------------<unsnip>-----------------------------------
You are correct. I had forgotten about SVS. I should have said "First 
release of IBM systems using Virtual Storage".

--------------------------------<snip>------------------------------------

> > The OS/360 Assembler F was named IEUASMxx. If you can find IEV90xx, 
> you'll have a version of
> > Assembler H that usually worked with OS/360 code and was MUCH faster
> > than IEUASM for system generation, expecially the Stage 1.
>
> Indeed. But all versions of Assembler H were IBM Program Products.
>
-------------------------------<unsnip>------------------------------------
No argument there. But for the amount of Assembler code in our shop, it 
was well worth the investment. :-)

--------------------------------<snip>---------------------------------------

>
> > Assembler G, from Waterloo was also a conciderable improvement over 
> IEUASM (Assembler
> > F) and ASMG is available from the CBTTAPE site.
>
> It's not clear if ASMG is any faster than Assembler XF (IFOX00). I'd
> love to know the politics behind IFOX; it appears to have been a
> re-implementation from scratch of IEUASM, to the identical specs, but
> with the minor additional requirement that it be reentrant. Why IBM
> would even think of doing such a thing, when Assembler H already
> existed, and was much more capable and much faster, I can't imagine.
>
------------------------------<unsnip>-------------------------------------
ASMG got most of its speed from the way it processed macros and maclibs. 
It read the SYSLIB directory into storage during initialization, which 
resulted in a LARGE speed boost over IEUASM. The advantage was in not 
doing BLDL or FIND for every macro in the code being assembled. In my 
experience (I must emphasize that phrase) it was quite a bit faster than 
IFOX00 as well, but suffered from device geometry differences in the 
SYSLIB concatenation. ASMG also had a "compile and execute" feature that 
was unique and highly useful for classes in Assembler Language.

IFOX00 had some enhancements to the macro language as well as the 
re-entrancy rewrite. Such trivial details as array dimension extensions 
and internal storage management (beyond reentrancy requirements), etc., 
but nothing earth-shattering. IIRC, there were some new variable symbols 
introduced, but I'm not sure about that. ASMH was still a "cash cow" at 
the time and the penny counters weren't willing to make any changes in 
the existing product structure vis-a-vis ASMH. Can we visualize the tail 
wagging the dog here? ASMG was much cheaper than ASMH and nearly as 
fast, so many shops chose the Waterloo offering over the IBM offering.

Rick