benedict@chaos.utexas.edu (Thomas Benedict) (05/30/91)
Say, I've been seeing several postings refering to IRIX 4.00 coming out someday. Sounds good. But specifically I was wondering if any of the following were intended to be stuck into 4.00: IRIX seems to have a problem keeping names straight, especially with regards to talk. At this point incorrect names seem to be more the norm than the exception. For instance, if I try to talk to someone else on the system they may see it as a request for a talk from someone other than me. On one occasion someone got a request to talk from himself. Another biggie on my list is the tape device drivers. I posted earlier about lost data because the nrtapens device driver skipped to end of tape and then rewound, causing a big loss of data. This is not only annoying, it's downright dangerous. The last entry on my 4.00 wish list is X11R4. We've got X terminals that will run R4, and even got a new update today. It hurts to be running these things at R3, especially when one of the neat features of R4 is that when an X terminal gets turned off it isn't the pain that it is under R3 to get it back up and running. Other than that, does anyone at SGI know when 4.00 will be released? Or if it's still unannounceable, just a rough guess at what year it will be out? :) Tom Benedict benedict@chaos.utexas.edu Center for Nonlinear Dynamics
russell@ccu1.aukuni.ac.nz (Russell J Fulton;ccc032u) (05/31/91)
benedict@chaos.utexas.edu (Thomas Benedict) writes: >Say, I've been seeing several postings refering to IRIX 4.00 coming >out someday. Sounds good. But specifically I was wondering if any of >the following were intended to be stuck into 4.00: >IRIX seems to have a problem keeping names straight, especially with >regards to talk. At this point incorrect names seem to be more the >norm than the exception. For instance, if I try to talk to someone >else on the system they may see it as a request for a talk from >someone other than me. On one occasion someone got a request to talk >from himself. This is a problem which is caused by processes not locking the utmp and wtmp files when they write to it. It causes the library routine cuserid to be unreliable on machines with lots of users. (The more users the more likely utmp and wtmp will get corrupt because of two processes writing to them at the same time.) This problem is not confined to Irix, its inherent in AT&T sysV. Some other vendors have fixed it and we are putting the heat on to get SGI to fix it as well. Maybe it will be fixed as result of the merge of SYSV and BSD? This bug also screws up connect time logging as done by the standard accounting software and sometimes even causes the accounting run to fail completely. Another particularly nasty effect is that when creating new users you use the option of letting the user set the password on first logon then sometimes the wrong password gets changed! Cheers, Russell. -- Russell Fulton, Computer Center, University of Auckland, New Zealand. <rj_fulton@aukuni.ac.nz>
olson@anchor.esd.sgi.com (Dave Olson) (06/01/91)
In <BENEDICT.91May30112819@chaos.utexas.edu> benedict@chaos.utexas.edu (Thomas Benedict) writes: | regards to talk. At this point incorrect names seem to be more the | norm than the exception. For instance, if I try to talk to someone | else on the system they may see it as a request for a talk from | someone other than me. On one occasion someone got a request to talk | from himself. As I recall (it isn't my area of expertise) there were some bugs with some programs not updating the /etc/*tmp files correctly, so that w and other programs (like talk) that read them sometimes got bogus info. I'm pretty sure this is fixed. | Another biggie on my list is the tape device drivers. I posted | earlier about lost data because the nrtapens device driver skipped to | end of tape and then rewound, causing a big loss of data. This is not | only annoying, it's downright dangerous. This only affected Exabyte, and only in a few cases. To the very best of my knowledge, this is now completely fixed. There were a number of changes to this driver to support some new devices and make it more configurable, in addition to the exabyte fixes. We also made some changes to reduce the amount of driver initiated positioning; look at 'man tps' when you get 4.0 for details. | The last entry on my 4.00 wish list is X11R4. We've got X terminals | that will run R4, and even got a new update today. It hurts to be | running these things at R3, especially when one of the neat features | of R4 is that when an X terminal gets turned off it isn't the pain | that it is under R3 to get it back up and running. Yes, X11R4 is part of 4.0 | Other than that, does anyone at SGI know when 4.00 will be released? | Or if it's still unannounceable, just a rough guess at what year it | will be out? :) Sometime in 91, at least so we hope :) -- Dave Olson Life would be so much easier if we could just look at the source code.