Jim Bohnsack <[EMAIL PROTECTED]> wrote: >I think that the biggest reason for keeping it in a DCSS is that it is >important code that you really don't want the casual user to muck around >with. How many problems have we all had to dig into only to find out that >there is a private copy of the XYZ EXEC on the user's a-disk that gets used >instead of the one that you've got out on a common disk. Maybe it would be >worthwhile to attempt to categorize execs in INSTSEG by frequency or >likelihood of usage, putting the seldom used ones all together so that they >will be in the page likely to be paged out. Even that, tho, given the >price and quantity of real storage kind of falls into the category of bit >twiddling. Do a cost benefit analysis and you'll probably find that even >the cost in your time of the analysis vs. the benefit of not having three, >4k blocks of SYSPROF exec in the INSTSEG, isn't there, let alone the real >cost of the storage the exec takes.
>This is similiar to what a marketing rep I used to work with said when he >would talk about the "grief to earnings" ratio of something. I agree 110%, Jim, but that won't stop an ObAnecdote: Back in the day at UofWaterloo, we ran a CMS monitor on the S- and Y-disks during the "slow" months of July and August that collected frequency of reference information for the EXECs and MODULEs on those disks. Then in our Copious Spare Time we'd try to convert those EXECs to MODULEs (this was before the Rexx compiler or INSTSEG) and/or make MODULEs re-entrant and put them in the nucleus. In maybe 1982, I was testing a local DIAGNOSE one afternoon that returned another user's MSG/EMSG/IMSG/SMSG settings. I wrote a tiny assembler program to test it. It was quick&dirty, so I didn't comment it or even do a REGEQU. Then I added the ability to read a list of userids to check. Then I added a few more features. Then someone put it on the Y-disk. You can guess where this is going: In the summer of 1985, when we ran that monitor, that program was the most heavily used program on the Y-disk. And it still had no comments or REGEQU, though it was over 2,000 lines of assembler by then. I decided it was time to leave and move to a different country before I had to fix something in it... ...phsiii P.S. I learned my lesson about commenting code, too, even if I today know lots of C/C++ programmers who say things like "You shouldn't comment your code because the comments might not reflect what the code actually does". Don't get me started...
