> > Consistency in using either the '-' or '_' is important. > > Yes, but possibly now too late. > > I recall seeing something somewhere that said that all HAL pins should > use "-" exclusively. > But then you have the problem that file names tend to use "_" (though > I just tested and "-" is allowed) and you end up with > loadrt some_hal_component creating pins called > some-hal-component.00.some-pin-name. > > But, if we change the pin names to be consistent now, every existing > config that uses the old names will break. >
Too bad. Then break them. I come from the world of strongly typed languages like ALGOL 60, ALGOL 68 and PASCAL so I don't really care if stricter rules break existing hard to understand or diagnose programs or config files. Back in the early days of computer languages that were done on punch cards I can remember our professor in our compiler construction course stressing that with the increase in the amount of memory available that the mantra 'add another pass' would become more important to create a list of error messages when compiling a program that were 100% accurate. He'd written an ALGOL-68 in IBM 370 Assembler that we used in the course. It was terribly processor hungry and since we were allocated only so many seconds of computer time compiling an assignment program too many times resulted in burning up our funds. At first that seemed stupid. Until we realized that fixing the first error statement and then recompiling was a waste of time. His 'add another pass' philosophy was such that if compiling a program generated 45 errors you'd better fix all 45 before trying to compile again. That's because unlike many other languages that have a flag that says stop after the first error this compiler recovered and kept going generating accurate errors. Not to say that after we fixed the 45 errors that new ones wouldn't show up. But once you experience that envronment, every other language that can't tell you everything that is wrong the first time is really just not as good. Programming languages of any sort that overload operators to be part of variable names but also math operators (even if only in other languages) is just plain stupid. I could go one but I also understand that arguing with the Linux community is like spitting into wind. So the end result of this is neither the program nor the documentation will change. And that's sad. John _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
