[Info-vax] Text processing examples with Fortran requested

[John Reagan]

"Michael Moroney" <moroney at world.std.spaamtrap.com> wrote in message 
news:hds5qo$heq$1 at pcls6.std.com...
> brooks at cuebid.usa.hp.nospam (Rob Brooks) writes:
>
>>moroney at world.std.spaamtrap.com (Michael Moroney) writes:
>
>>> I have to deal with thousands of lines of the most gawdawful macro-32
>>> code ever created.
>
>>yeah, well, you spent time in SHDRIVER, so you should be well-suited
>>to the task of dealing with convoluted MACRO-32 :-)
>
> While SHDRIVER was certainly convoluted, it's _nothing_ compared to the
> hideousness of what I'm dealing with now.  Seriously.

Then I'm very afraid for you!

During the port to Itanium, we instrumented the IMACRO compiler to print out 
various internal statistics in the .LIS file (near the command qualifier 
summary).  One of them was the total number of flow blocks found by the 
compiler's flow analyzer.  This, along with the number of entry/exit points; 
routines jumping into each other; etc. was used to compute a "complexity" 
factor.  SHDRIVER blew everybody else out of the water.  By far the most 
complex piece of Macro-32 code in the entire system from a compiler analysis 
point of view.

Other Macro-32 code like RMS and DCL is complex with having all sorts of 
global data in registers which mutates slightly as the code branches 
into/outof various routines, but the compiler in general doesn't care about 
those (other than nasty code to deal with NaTs that might have leaked into 
the Macro-32 code).

John