karl@ddsw1.MCS.COM (Karl Denninger) (12/07/88)
Hello Net.Oracle! We have a small problem. SCO Xenix has been very good to us until now; but I am afraid we have hit a brick wall somewhere in their 286 compiler or libraries. Environment: Large integrated package (conferencing software). Executes unchanged (source-wise) on SCO V/386, Microport V/AT and other SVID-complient systems. On SCO Xenix, with the '286 compiler, everything appears "ok" until you enter the portion of the package where curses is used. In large model you then get a segmentation violation on the "initscr()" call! Built with huge model (yep, it even builds that way) I get normal operation -- as long as I don't try to open a second window. Opening a second window brings yet another segv! (this time on "newwin()"). Since both of these are in library routines, and we don't have source to them, it appears we are in deep dodo. We're set up for terminfo curses at the moment. Under termcap curses everything works until you try to wrefresh(window) the second window; then you ALSO get a segmentation violation. Termcap curses is a bad deal for this application anyways, given that it doesn't support the keypad.... Other packages using curses that we have here compile and run properly. To the best of our knowledge the only differences here are: 1) AKCS is a _large_ program; we needed to use the following the compiler directives before we could get a successful link: -M2let1000 -SEG 256 2) AKCS mallocs several buffers when it starts up and only calls initscr() when a user requests a windowed environment (which may be later or never); all of our other packages do the "initscr" as a "first thing in" operation. Help SCO (or anyone else with ideas!); we're going nuts out here! -- Karl Denninger (karl@ddsw1.MCS.COM, ddsw1!karl) Data: [+1 312 566-8912], Voice: [+1 312 566-8910] Macro Computer Solutions, Inc. "Quality solutions at a fair price"