Bakin@MIT-MULTICS.ARPA ("David S. Bakin") (08/06/86)
OK, this is just to let you know that the mail is piling up around here on my question on use-clauses, with the current sentiment being not to use them at all. I'll summarize next week, but in the meantime, anyone who sends me mail saying not to use use-clauses at all please include a description of how you access user-defined operators. Not how you should, how you actually do. By the way, I don't buy the arguments about code with use-clauses being harder to read. 1) You should be using an interactive tool to help you understand your code (more about this in a later message, if anyone jumps at the bait) 2) I find it harder to read code that stretches across the page and down across pages because of full qualification. Anyway, keep those cards and letters coming, folks! (For those of you who DO use use-clauses, please let me know how your compiler fares with them! At least one compiler spec sheet that I've seen says it "processes withed packages at several thousand lines per minute" which of course implies that it takes time to scan withed packages instead of just having direct access in a library. Does anyone use a compiler whose time-to-compile grows based on the size of the withed and/or used context? And if so, how does it affect your programming style?) -- Dave (Bakin -at mit-multics)
alden%jade@spp1.UUCP (08/07/86)
David,
You can access operators with prefix notation and dot (.) notation with the
package name.
my_var := my_package."+"(op1,op2);
Also, the arguement for using use based on long names is still weak. If
you have a problem with long package names then use RENAMES, again as local
to the block affected as possible. Renames can be used to abrivate the
package name or to abreviate the package and operation combination.
The arguement for tools to analyze code is a good one. To date, the only good
tool to do this kind of analysis is on the Rational R1000 computer. Have
you got $750,000.00 to give me so that I can use the tool?
... Tony Alden
TRW