keld@diku.UUCP (Keld Simonsen) (06/13/84)
Now I have for God knows which time installed new uucp, news ... software on our different machines, pdp, vax, m68k's. And I know this is going to happen many more times in the future... I have used about a week now, finding small bugs, tailoring the system, testing, making the same corrections to the source to specify machine type, OS type, machine name etc. I think it would have saved me quite some time if the old (pre 2.9 news) way of doing it via /usr/include/whoami.h definitions had been used. /use/include/whoami.h is an obvious place to put information on the system, including name of organisation, address etc., items which are not provided by USG uname(2) and BSD functions, but needed e.g. in news. This of cause requires the installer to be root/bin, but you have to be privileged anyway to install things like uucp and news, and it is only once that the system administrator has to do the <whoami.h> file. I do not see that this scheme could give problems to binary distribution these problems are there allready, and they might be eliminated by writing something like the uname.c distributed with news for getting long name of the organisation, etc. Not to mention, a lot of other software systems may benefit from a <whoami.h> file too. A problem is what to be defined, and how. E.g. Unix versions have quite some names for the same level in different software. V6 (a nice little system), V7, PWB are well establised. But what about SIII/S3, SV/S5, SVRII/S5R2, BSD41C/BSD42 etc? And what about common denominators like USG and BSD ? What is to be considered a common denominator, I have seen BSD41C used as indication of a 4.2 system, they have the same directory structure ! I think this problem should be addressed even if you do not like the <whoami.h> solution, it would be nice to have unique names for the systems. I prefer the S3/S5/S5R2 as shorter and more readable. I think we should define a minimum set of names in <whoami.h> and the associated appropiate values. There should be what uname(2) now gives us, a MYORG, maybe a USG or BSD entry, and a PROCESSOR and a MACHINE def. Keld Simonsen, Institute of Datalogy, University of Copenhagen mcvax!diku!keld
keld@diku.UUCP (Keld J|rn Simonsen) (07/05/84)
I posted this some three weeks ago, when everybody else were in SLC and the connections between Europe and North America were not that good. I have seen no responses on it, so here it goes again: Now I have for God knows which time installed new uucp, news ... software on our different machines, pdp, vax, m68k's. And I know this is going to happen many more times in the future... I have used about a week now, finding small bugs, tailoring the system, testing, making the same corrections to the source to specify machine type, OS type, machine name etc. I think it would have saved me quite some time if the old (pre 2.9 news) way of doing it via /usr/include/whoami.h definitions had been used. /use/include/whoami.h is an obvious place to put information on the system, including name of organisation, address etc., items which are not provided by USG uname(2) and BSD functions, but needed e.g. in news. This of cause requires the installer to be root/bin, but you have to be privileged anyway to install things like uucp and news, and it is only once that the system administrator has to do the <whoami.h> file. I do not see that this scheme could give problems to binary distribution these problems are there allready, and they might be eliminated by writing something like the uname.c distributed with news for getting long name of the organisation, etc. Not to mention, a lot of other software systems may benefit from a <whoami.h> file too. A problem is what to be defined, and how. E.g. Unix versions have quite some names for the same level in different software. V6 (a nice little system), V7, PWB are well establised. But what about SIII/S3, SV/S5, SVRII/S5R2, BSD41C/BSD42 etc? And what about common denominators like USG and BSD ? What is to be considered a common denominator, I have seen BSD41C used as indication of a 4.2 system, they have the same directory structure ! I think this problem should be addressed even if you do not like the <whoami.h> solution, it would be nice to have unique names for the systems. I prefer the S3/S5/S5R2 as shorter and more readable. I think we should define a minimum set of names in <whoami.h> and the associated appropiate values. There should be what uname(2) now gives us, a MYORG, maybe a USG or BSD entry, and a PROCESSOR and a MACHINE def. Keld Simonsen, Institute of Datalogy, University of Copenhagen mcvax!diku!keld