goldberg@su-russell.ARPA (Jeffrey Goldberg) (01/09/87)
Basically my question is is there a work around to the fact that under 4.3 some programs (vi and more in particular) don't inspect the termcap entry to determine the number of lines or the number of columns of a given terminal. I recently got an account on a 4.3 machine (a VAX 780) and I moved some stuff over from a machine running 4.2. What I have is my own little termcap file that contains the entry for an hp9836 which I found on one of the local Vaxen and works fine, and my own thing which is just a 36 with 40 lines instead of 24. This works fine on the 4.2 machine, but doesn't on the 4.3 machine. Vi and more behave as if I were working on a 24 line terminal. Looking at the manual page for termcap on 4.3 I find under BUGS and CAVEATS: "Lines and columns are now stored in the kernel as well as in the termcap entry. Most programs now use the kernel information primarily; the information in this file is used only if the kernel does not have any information." I don't pretend to begin to understand where the kernel gets its information about these terminals, but there is no indication that anyone has ever defined a terminal with the name that I give it. So I don't understand how the kernel could have prior knowledge of it. I do use tc extensively in my entries. Should I just duplicate the stuff in different entries instead of using tc? Jeff Goldberg ARPA: goldberg@russell.stanford.edu, goldberg@csli.stanford.edu UUCP: ...!hplabs!russell.stanford.edu!goldberg cryptography, terrorist, DES, drugs, cipher, secret, decode, NSA, CIA, NRO. The above is food for the NSA line eater. Add it to your .signature and you too can help overflow the NSA's ability to scan all traffic going in or out of the USA looking for "significant" words. (I am told that this is no joke.)
gwyn@brl-smoke.ARPA (Doug Gwyn ) (01/09/87)
In article <234@su-russell.ARPA> goldberg@su-russell.UUCP (Jeffrey Goldberg) writes: >"Lines and columns are now stored in the kernel as well as in the >termcap entry. Most programs now use the kernel information >primarily; the information in this file is used only if the kernel >does not have any information." The 4.3BSD kernel (following the lead of Sun and/or AT&T) maintains information about the dimensions of each "terminal" (which may actually be just a window on a multi-window display) in a /* * Window size structure */ struct winsize { unsigned short ws_row, ws_col; /* character size of window */ unsigned short ws_xpixel, ws_ypixel; /* pixel size of window */ }; That contains all 0 data if the "terminal" size has never been set. The only way to set the size is via a TIOCSWINSZ ioctl, and the only way to query the size is via a TIOCGWINSZ ioctl, both of which take a (struct winsize *) as their third parameter. (Note: DMD-supporting UNIX System Vs, such as SVR3, have analogous information contained in a (struct jwinsize) with a JWINSIZE ioctl to retrieve it; it's updated automatically inside the kernel when a window size changes.) The general procedure for determining a terminal's screen size is: 1) Try a TIOCGWINSZ. If it fails, or if it returns 0 data, then proceed to step 3; otherwise use the returned information as the terminal's screen size. 2) Some sites have added pagination to their terminal handler; in this case one might be able to obtain the number of rows via an ioctl analogous to the procedure of step 1; if supported and successful (and non-zero!), use this value and proceed to step 3 just to get columns information. 3) Try to fetch COLUMNS and/or LINES from the environment (if not yet known from previous steps). If either is available, use it for the terminal size in that direction; if either is not set in the environment, find it out by proceeding to step 4. 4) Get TERM from the environment and use it in a tgetent (or equivalent call to initialize the display-handling package). The termcap capability "li" (terminfo "lines") is the default number of lines and termcap "co" (terminfo "cols") is the default number of columns to assume for that terminal type. 5) If one still hasn't determined the terminal screen size, assume that it is an "infinite quadrant". This probably isn't correct, but then it's what the user deserves for not setting up any information about his terminal.
colonel@sunybcs.UUCP (01/11/87)
I don't know about the kernel, but a lot of programs look at the
environment variable $termcap, which is set at login. You can reset
it with "setenv".
--
Col. G. L. Sicherman
UU: ...{rocksvax|decvax}!sunybcs!colonel
CS: colonel@buffalo-cs
BI: colonel@sunybcs, csdsiche@ubvmsjc@piaget.UUCP (John Cornelius) (01/19/87)
In article <234@su-russell.ARPA> goldberg@su-russell.UUCP (Jeffrey Goldberg) writes: >Basically my question is is there a work around to the fact that >under 4.3 some programs (vi and more in particular) don't inspect >the termcap entry to determine the number of lines or the number of >columns of a given terminal. > >I recently got an account on a 4.3 machine (a VAX 780) and I moved >some stuff over from a machine running 4.2. What I have is my own >little termcap file that contains the entry for an hp9836 which I >found on one of the local Vaxen and works fine, and my own thing >which is just a 36 with 40 lines instead of 24. This works fine on >the 4.2 machine, but doesn't on the 4.3 machine. Vi and more >behave as if I were working on a 24 line terminal. > >Looking at the manual page for termcap on 4.3 I find under BUGS and >CAVEATS: > >"Lines and columns are now stored in the kernel as well as in the >termcap entry. Most programs now use the kernel information >primarily; the information in this file is used only if the kernel >does not have any information." > >I don't pretend to begin to understand where the kernel gets its >information about these terminals, but there is no indication that >anyone has ever defined a terminal with the name that I give it. >So I don't understand how the kernel could have prior knowledge of >it. The command "stty rows #" where # is the number of rows causes vi and more to work correctly. I believe the same is true of columns. The command "stty everything" will show you the current settings. See stty(I). >I do use tc extensively in my entries. Should I just duplicate the >stuff in different entries instead of using tc? > >Jeff Goldberg >ARPA: goldberg@russell.stanford.edu, goldberg@csli.stanford.edu >UUCP: ...!hplabs!russell.stanford.edu!goldberg > >cryptography, terrorist, DES, drugs, cipher, secret, decode, >NSA, CIA, NRO. > The above is food for the NSA line eater. Add it to your > .signature and you too can help overflow the NSA's ability to > scan all traffic going in or out of the USA looking for > "significant" words. (I am told that this is no joke.) -- John Cornelius (...!sdcsvax!piaget!jc)