eric@CAM.UNISYS.COM (Eric Wels) (01/20/88)
We keep getting IOT trap errors! These are very annoying since NeWS aborts everytime we get one of these errors. We haven't been able to pinpoint a specific cause of the errors but they frequently occur when: a pop-up menu is envoked, a dbgwhere is executed after a dbgenter (zombie & wait), while reading in 'good' code (code which normally works), and 'zapping' windows (killprocess group). The IOT trap errors also occur at other times but typically happen in 'groups'. That is, once IOT trap errors begin, they will continue every 10 minutes to an hour for the rest of the afternoon. The very next day, running the same code, they may not happen at all. Our code seems to be more prone to these errors than the code supplied with NeWS -- however, we've seen IOT trap errors while using the icondemo program. OK -- does anyone out there have a list of conditions which NeWS responds to with an IOT trap signal? Is anyone else experiencing these types of problems? Does anyone have an educated guess as to what might be happening? Does anyone know how to reclaim a canvas? Thanks bunch. P.S. We still think highly of NeWS. arpa eric@CAM.UNISYS.COM These views are mine and do not reflect an offical position of my company.
guy@gorodish.Sun.COM (Guy Harris) (01/20/88)
> OK -- does anyone out there have a list of conditions which > NeWS responds to with an IOT trap signal? IOT traps, on most UNIX systems, are generated by the "abort" library routine. In SunOS 3.2 and later, "abort" generates SIGIOT. Unfortunately, I don't know what would provoke NeWS to call "abort"; there appear to be a number of places in the NeWS server source where "abort" is called, including something that might well be a PostScript operator permitting PostScript code to generate an "abort".... Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.com