Re: Long names are doom ?

[Gary Scott <scottg@flash.net>]

Hi,

It is possible to write highly mnemonic names with 6 characters.  I
agree it's too short, but consistency of design helps tremendously and a
quick study of the documentation for the library itself coupled with a
general knowledge of the subject matter makes it reasonably evident:

GSLT = graphics subroutine line <set> type
GSQLT = graphics subroutine query <get> line type
GSLW = graphics subroutine line width
GSQLW = graphics subroutine query line width

I have probably 500-1000 routines of this type virtually memorized
(between GDDM and GINO) with many of the parameter types and value
ranges memorized as well.  Once you've been programming this way for 20
years, it isn't difficult at all to read the code.  Obviously not easy
for a novice or someone unfamiliar with the libraries to do so.  There
are few if any users of Fortran that ever thought 6 characters was
enough and nearly all compilers allowed more.  Some compilers had the
abhorent habit of ignoring characters beyond 6, however.  

31 is a reasonable minimum in my opinion, but it was put into effect
about a decade ago.  If someone wants to change it, formally approach
the standard committee with a proposal.



Shankar Unni wrote:
> 
> AG wrote:
> 
> > I once saw some code which used (in the same namespace) identifiers
> > PassThruID  and PassThroughIDs. They were apparently introduced
> > by separate programmers over time and worked fine until someone
> > confused them. Took ages to sort it out. So, while those two aren't
> > even at 31 char limit, one could argue that 6 char limit would have
> > precluded that sort of error :)
> 
> Really?  Would it have been somehow been a lot clearer, and harder to
> confuse the two, if they had been, say, "PTID" and "PTIDS"? I'd think
> the opposite, actually..
> 
> You're right in that infinitely long names don't help beyond a certain
> point, but 6 chars is ridiculous. Forcing ultra-short names has the
> effect of people using an acronym soup for variable names, and "ABCXQZ"
> is pretty easy to confuse with "ABCQXZ".
> 
> Here's an *actual* example of code I saw posted in another newsgroup
> (names erased to protect the guilty - or in this case, the innocent!).
> And these are 12-char names - think how much more readable this would
> have been with 6-char names:
> 
> >                 double sicjpcisj=*scratchptr;
> >                 double sicjncisj=*(scratchptr+1);
> >                 double cicjpsisj=*(scratchptr+2);
> >                 double cicjnsisj=*(scratchptr+3);
> >                 double qcksicjpcisj,qcksicjncisj;
> >                 double qsksicjpcisj,qsksicjncisj;
> >                 double qckcicjpsisj,qckcicjnsisj;
> >                 double qskcicjpsisj,qskcicjnsisj;
> 
> --
> Shankar.


-- 

Gary Scott
mailto:scottg@flash.net

mailto:webmaster@fortranlib.com
http://www.fortranlib.com

Support the GNU Fortran G95 Project:  http://g95.sourceforge.net