1.0.9 --> 1.10 fixed realloc() for suns. *sigh*. realloc(0, n) fails on a sun. replaced fprintf() with debug_ll(). fixed language in a few places. correction: alpha/osf1 support is by Craig I. Hagen, not marius hancu. My mistake. apologies. Bug introduced into upload fixed. Was causing all transfers to fail leaving zero length files. Delay that was inadvertently left in removed. There was a 1/10th of second delay in the main loop, resuling is awful thruput on v.high speed lines. If the baudrate is set to 0 (i.e. -s0 ) then rate limit is not done. This is useful on higher speed links. The normal window/timeout mechinism is more than good enough on links > 9600. Ps. This is not a version of term. See 109 release notes. I'm not sure quite what it IS, but is definately isn't a version. :-) Added the delay back in, but made it smaller. It drops the CPU time used by an order of magnitude.... 1.0.8 --> 1.0.9 3 fixes by Kay Roemer. Term will now close some files that it wasn't before, read buffered data from a child that has exited, and close the right sockets for children. Alpha/osf1 support from Marius Hancu Prevent compressing more than 2K of data into a single packet. The biggy. Preventing data overruns. People using TCP connections that have very compressable data can now work in peace. Term previouly used fixed size buffers. Now it uses dynamic buffers. This means that if a client is a bit slow in reading data from term, term will increase the buffer size instead of throwing away data. On the down size, term's memory usage is now theoretically unlimited. This should fix most if not all of the problems people had with X over term. I finally got bitten by this once too often. THIS is DEFINATELY the last version of term. I hope. 1.0.7 --> 1.0.8 Well, I did what I swore not to. Here's a new version of term. Fixed a bug relating to overflow on a timer variable. Caused term to stop working sometime on Oct 26th.... *blush*. Jeff and I found this at the same time (probably along with a LOT of other people.. :), and I used jeff's patch. (cleaner than mine). Merged in olaf's patches. Olaf, I've managed to delete your email. Please mail me again. See 'Blurb' for olafs changes. I suppose I'll have to support this bloody version too! argh!!!!. :) --------------------------------------------------- 0.3--> 0.4 Fixed upload to make the optinal arg really optional. Fixed main.c to NOT cat the file arg to the socket name. 0.4--> 0.5 Added baudrate option + env variable. Xconn now checks the DISPLAY variable to know what display to use. Fixed bug in xconn relating to printf. Fixed main.c. Was useing wrong lower bound for baudrate. Thanks to Someone whoes name escapes me. amil me again pls. 0.5--> 0.5.1 Silly bug in strncmp() in main.c 0.5.1.--> 0.5.2 Fixed a few out of date docs. 0.5.2.--> 0.5.3 The argument parseing got improved a lot. Thanks to M. Saggaf. 0.5.3 --> 0.6 Xconn got modifed to use an INET socket instead of a UNIX socket. This means you can run term on one machine, and use x apps on another. 0.6 --> 0.6.1 I wrote a Q&D hack to see how the serial line performs. Consists of checkline.c and linerem.c (for local and remote ends). See readme for details. 0.6.1-->0.6.2 Fixed up pty.c a bit more. Hopefully it should the child proc should be able to lose the controlling term on a sun now. :( There was a damned silly bug. I had forgotten to #include config.h Fixed xconn to properly print the display it was connecting to. Fixed xconn to look at DISPLAY if it existed and use that for the display number. Fixed up sun stuff a bit (EAGAIN instead of EWOULDBLOCK). Fixed up terminal.c to set VMIN and VTIME 0.6.2 -> 0.6.3 Added a ~/.term/termrc Should be able to set everything from the termrc file, includeing escaped characters. See TERMRC for format. Beware, it isnt tested. :) Also, at the moment, it assumes that adding 32 to an escaped character, won't produce another escaped character. Two new features. a -f option tell term to generate control-Q's ( in case line noise generates a control-S ). In a termrc you can specify a 'shift'. This simply get XOR'ed with all character before they are sent, and after they are received. The purpose being to map a lesser used range on characters down to 0-32 (which are all escaped). I use 224 (0xE0). Haven't measured the speed up. But histogram of characters received shows a drop in the number of escapes sent. 0.6.3 --> 0.6.4 I was forgetting to close the file descriptor used to read the termrc file. 0.6.4 --> 0.6.5 A goddam silly bug. was doing if(strcmp) instead of if (!strcmp) SIGH!!!!!!!!! 0.6.5 --> 0.6.6 Another silly one. Wasn't escapeing characters properly. 0.6.6 --> 0.6.7 Changes the way the input window was worked out. Hopefully it fixed somes bugs reported by some ppl. 0.6.7 --> 0.7.0 Re-wrote the compression tree handleing. Hopefully a lot clearer, and understandable. Hopefully less bugs too. Fixed a few of the include files to remove warnings. Fixed a a bug in upload.c Was doing i == 0, instead of i =0. Fixed a bug in test.c had arguments in wrong order. Thanks to fjh?? for pointing this out. Fixed serial.c so that five 0's will exit term regard less of what the byte_shift is set to. 0.7.0 --> 0.7.1 Added support for SGI. Thanks to 'warwick' harvey. Fixed xconn to properly look at DISPLAY, search for the first unused display. The way to run it now is... "setenv DISPLAY=`xconn`" Reduced window size to 2. Don'T THINK it will adversly affect throughput but haven't extensively tested it. 0.7.1 --> 0.7.2 Added options to configure both the window size, and the paacket timeout values. Changed the default window size up to 3. A size of 2 at 2400 baud introduces too much latency into keystrokes. Added an option to toggle transmit forceing. Forceing is now off by default. Not tested properly. Added the -w , -t flags. 0.7.2 --> 0.7.3 Fixed bug in xconn relateing to searching for unused display. Added the -n flag. See OPTION, and TERMRC for details. See also the source. :) Cleaned up some more prototypes. Compiles are now cleaner. 0.7.3 --> 0.7.4 Added handleing of escaped characters that produce another char that should be escaped. 0.7.4 --> 0.7.5 Fixed noise echoing to remove byte shifting. Tried to write some BSD terminal handleing. Someone who has BSD want to test this?? The famous compression bug found!!!! The all-singing-all-dancing- wonder is gone! Many many thanks to Janne Sinkkonen (sinkkone@ cc.helsinki.fi) for this. I was clearing 300 bytes , on a 256 byte buffer. :( 0.7.5 --> 0.7.6 More prototype fixes. Removed the call to abs() in link.c (convex does have it apperently) Fixed checkline.c to send an XON after every character (so it doesn't 'freeze'), and increased rate to 16 characters per second instead of 1. There is a new client 'redir'. Usage is 'redir '. So you can do 'redir 23 4000' and then everyone who telnets to port 4000 on your remote machine, will get connected to telnetd on your local machine. 0.7.6 --> 0.7.7 Fixed a bug in main.c. Fudge_flow was supposed to be a counter limit, not a flag. 0.7.7 --> 0.7.8 Fixed makefile. 'make test' make the wrong file. Changed the way error handling was done. Added in LL read/write debugging. Added support for ranges in termrc. i.e. 'escape 0-32' should work. Not extensively tested. Lots of structural changes internal to term. Missing break; in main.c -n was broken. fixed now. An incredibly stupid bug in serial.c fixed. I was useing && instead of ||. Resulted in the retransmition being braindead. *SIGH!!!!!* Resulted in the packet window being as full as possible instead of the reverse. Changed the algorithm used to decide re-transmition. Now it will never timeout a packet before the 'timeout' value. Will not write out an error message for any line of the ~/.term/termrc file that it doesn't recognize. Redir can now redirect to a host. I.e. 'redir 119 my.news.server:119' on the linux box, lets you treat you linux box as an NNTP server. Thanks to Rick Sladkey. Renamed all the clients to get a 't' in front. term --> term rsh --> trshell upload --> tupload xconn --> txconn rsystem --> trcmd redir --> tredir 0.7.8 --> 0.8.0 Replaced waitpid() with wait3(). Added support for NeXT Added support for AIX Added (partial) support for BSD Thanks to cruncy!croutons@dartmouth.edu we have support for individually compressed streams. I.e. you can be uploading a file with no compression, while running a shell whoes output does get compressed. Added priorities for streams. Vastly improved responce time with multiple streams (I hope.. ). Achieved by checking for new clients ready more often. The old scheme was to take data from a client untill it had none left before looking for a new client. Now it looks for a new client at the beginning of every packet. This is what it should have been doing anyway. Re-wrote the client option handleling. Now all clients take 4 arguments -t Tries to connect to the socket ~/.term/socket Default is "". Useful only if you are running multiple terms (over multiple serial connections). -r Turns off compression for this link. -c Turns on compression for this link. the above two options affect compression both ways. The capacity is there to only compress one-way. Tell me if you want this? -p Changes the priority of the link to . Default is zero. tshell has a default of 2. tupload has a default of -2. The priority is currently absolute. i.e. if a there is a client with a high priority with data to send, then no client with a lower priority will send untill the higher has finished. This probably needs improve- ing. 0.8.0 --> 0.8.1 Redid all the #include files. Simplified (hopefully), and consider- ably easier to port. Fixed a bug in link.c. Was very silly. It could end up switching channels in the middle of a control sequence. Added signal handleing for kill signals and segv's. 0.8.1 --> 0.8.2 Fixed another silly bug in 'tredir' client. Was introduced in version 0.8.0 Fixed bug in makefile. Had missed a few places that were useing the old names of the clients. Fixed bug in xconn.c Was useing wrong filename for the X-window unix-domain socket. Added some more debugging calls. -d64 should be fairly informative for control info now. Added quick+dirty 7 bit support. run term with '-a' on both ends to use it (a == seven. maybe.). Or put "sevenbit" in your ~/.term/termrc file. WARNING!!!!!! It is very rudimentry. In particular, unless you send everything compressed, then it will strip the top bit of it. i.e. uploading file when not useing compression with corrupt them. WARNING 2!!!!!. It is in no way tested. Mail me how it goes.. Makefile got extensively re-written. Thanks to croutons (again :). 0.8.2 --> 0.8.3 0.8.2 was basically pretty stuffed try again. Fixed bug in compress.c (introduced in 0.8.2). Fixed a few minor bugs in link.c relating to put_client_data being called with an un-initalized structure. Hopefully fixed the bug with things not exiting when they quit. Really fixed the above bug. Telnet through a redir will now exit properly. (was a bug related to the changed format for control sequences. Read PROTOCOL.unix for details). Brought PROTOCOL.unix up to date. 0.8.3 --> 0.8.4 Fixed prototype spelling error in link.c Tried to fix trcmd. Does anyone use this?? Fixed a few race conditions WRT control messages. Was clearing buffers at some in-approrite points. Should fix the multiple upload bugs. Fixed get_next_client so that clients with the same priority share the line equally. 0.8.4 --> 0.8.5 Fixed the bug that was corrupting X windows streams. Fixed a (potenial) bug in redir.c (Same one as xconn.c). Removed the trcmd client. Renamed the trshell client to be 'trsh'. 0.8.5 --> 0.8.6 Removed the C_SYSTEM command, and associated files. Fixed a bug in xconn.c WRT searching for an un-used display. Thanks to croutons again, trsh now takes multiple arguments happily. commands like 'echo this is a multiple argument command' should work just fine. Wrote a CREDITS file. Anyone I have missed please mail me. Applied Rick Sladkey's changes so that C_SOCKET will work with a dotted decimal address. 0.8.6 --> 0.9.0 Removed 7 occurences of dubious language. Fixed pty.c to test for more pty's. Thanks to Rick sladkey for pointing this out. It now tests pty{p,q,r,s,t,u,v,w,x,y,z} if this isn't enough, let me know, and I will fix it for more. Went and implemented return codes for all commands. This means that you shouldn't get any 'hangs' when attempting to use non-existant ports and other such oddities. This also changes the command protocol which means you have to upgrade both ends, and all your clients to use it. tupload will now do recovery. If you try to upload a file that already exists on the remote end, then term will seek to then end of the file, and to the equivilant position on the remote file, and start uploading from there. Not fully tested. 0.9.0 --> 0.9.1 Write a minimal 'tstat' client. It currently just tells the compression ratio's that the local and remote ends have achieved. Write a Q&D strerror() for sun's. It is in lib.c Fixed a small bug in lib.c. Was doing if t<0 && !t ... 0.9.1 --> 0.9.2 We finally have some genuine man pages! Thanks very much to gtaylor (see CREDITS file). Added support for HPUX. Added in a new checksum routine. This one is a genuine CRC. Started to merge checkline into term proper. Changed most things to use unsigned chars. Should fix a few problems on sun's. Eliminated a race condition in pty.c 0.9.2 --> 0.9.3 Fixed a small bug in man pages. Updated the README file. 0.9.3 --> 0.9.4 Updated the escapeing mechinism. It is now dependant on the direction of the link. There are 3 new termrc commands. "escape" now simply tells term never to transmit that character. "ignore" tells term to silently strip that character from the input stream. I.e. if your modem uses software flow control use "ignore 17" and "ignore 19". "seven_in" tells term that the remote end is seven 7 bit tokens. "seven_out" tells term to transmit seven bit tokens. So in the unlikely event that you have a seven bit link TO to remote host, and an 8 bit link FROM the host. Lets say that it chews all control characters and 126. The 8 bit link is clean. No characters get chewed. Your local termrc should look like.. seven_out escape 0-31 escape 126 and the remote termrc. seven_in ignore 0-31 ignore 126 "sevenbit" now is the same as doing both "seven_in" and "seven_out". 0.9.4 --> 0.9.5 Fixed silly bug WRT seven bit mode. Was sending packet that were too long. Added (a little) more into to tstat client. 0.9.5 --> 0.9.6 Added an error check to tredir, to check for errors in binding the local socket. Fixed the sevenbit support a bit. It was still possible for term to try to generate packets longer than 127 bytes. 0.9.6 --> 0.9.7 Fixed compression for seven bit mode. 0.9.7 --> 0.9.8 Added more info to C_STATS Change client.c to include varargs.h for a sun. Mail me if this is wrong. Changed 'server' to be 'term_server' to avoid (potenial) conflict when linked with other programs. Fixed a problem in client.h whereby un_char might not have been defined. Fixed the man pages WRT tredir %d %s:%d. Added tmon client. Fixed a few minor bugs in the same. Extensive internal changes. Everything now uses the Buffer struct, and all the oddball buffer routines have been cleaned up. Also, partial support for bound sockets has beed added. Added support for SIGWINCH. Written by Teemu Rantanen, cleaned up slightly by me. Doesn't (yet) handle on-the-fly resizes. 0.9.8 --> 0.9.9 Fixed an include file bug in trshell.c Wasn't #include'ing signal.h Implemented a remote bind protocol. See the two new sever commands C_BIND, and C_ACCEPT. Fixed some more char/un_char conflicts. Fixed tmon.c (wasn't includeing sys/time.h). A new and improved version of tmon is included. This one is much more accurate. 0.9.9 --> 0.9.9a Fixed a HZ in tmon. Should compile most places now. Half fixed a bug relating tmon.c. It works, but there is still a race conditon. The problem is if two clients start at opposite ends at the same time, they wil get the same number. This is bad. (Hopefully) fixed SGI support. Tried to fix a condition relating to EWOUDLBLOCK. (Could people please try and let me know how to reproduce a bug if they see it!?) 99b Fixed all the check_client() stuff. Should exit more cleanly now. *cross fingers*. 99c Improved the Makefile, so there is only one refernce to the compiler. Fixed the select(). It nows passed the number of fd's to check, not just 64. Added support for varargs.h. varargs is used instead of stdargs if the #define USE_VARARGS is defined. Fixed spelling of 'compiling' in README file. (Thanks to cfs@nir.gatech.edu [Charles Stevens]). Changed xconn.c to do a little more error checking. Improved the README to be a little clearer. 099d Documented the commands in client.h Added a C_DUMP command. usage is C_DUMP %d, meaning to go dumb for the next %d bytes. See protocol.unix. Removed the C_CLIENTNUM, and replace it with a C_STATS sub-function. Documented all the new commands in PROTOCOL.unix. Happy now Quarters? :) Added in Jeff Grills new upload. Fixed tmon to remove all references to termio{s} from it. Added calls to terminal_raw() so that any key will imediately exit it. Read the key before exiting to ensure it doesn't get read by the shell. Fixed JG's new upload so that it uses C_DUMP, instead of opening a new connection for each file. Aded a C_STAT at the end to ensure that tupload doesn't exit before the entire file is there. Fixed read_into_buff to take the size. (needed for new upload). 099e Fixed a few bugs in upload. 099f various bug fixes and code cleanups. (read: I can't remember what I changed. :) Added a bit of sanity checking in link.c:get_client_data(); 099g More sanity checking in read_into_buff/write_from_buff(); Added a bit more information to tmon. Extended C_STAT to read file types and permissions. Re-wrote much of xconn.c to be more robust, and smaller. Re-wrote xconn a bit more, and redir as well. Moves the bulk of the (common) code to connect.c. 099h Updated PROTOCOL.unix Added S_IFREG, and S_IFDIR to link.c for those machines that don't have them. (i.e. F****** suns!). Changed the dependancies to NOT include the system header files. Tried to get term to compile nicely on a 386BSD. Fixed Makefile.{sun,sgi} to include connect.c Fixed xconn to print out the display number it is useing. export DISPLAY=`txconn` should now work. Made tredir move itself into the background automatically. No longer need to invoke it with 'tredir ... &'. Replaced a few 'char' with 'un_char's. There was one in serial.c that was screwing compilers that used signed chars. Upgraded upload to match Jeff Grills. The new upload has lots of features that aren't yet documented. Improved pty.c and general hpux support. Thanks to H.J. Lu Re-wrote bits and pieces of upload to suit myself. Made gratuitous changes to the formatting on the way. 099i Removed the automatic escapeing of characters < 32. Now escapes NO characters be default. BEWARE!!!!!! If this causes you problems, a quick fix is to add 'escape 0-31' to your ~/.term/termrc file. This improved my CPS rate by some 12% just up escape 30 instead of 32 characters. Added a signal() to ignore SIGPIPE's. This was what was killing term at in-opertune times. Fixed a bug in C_STAT relateing to permissions. Made tupload use no compression by default. If you want to upload something and have it be compressed over the link, you will have to use 'upload -c ...'. Term can't handle compressing stuff that is already compressed. Txconn, and Tredir will now only run when term is running, and will exit when term exits. Tell me how you feel about this change. 099j Tried to clean up shell exiting. Fixed up the pty.c code. It was checking for the wrong pty's first. The patches hpux had broken pty's for other machines. 099k Prevented term from sending five '0' itself accidently, and killing the remote term as a result. Moved the detection Fixed S_ISREG's in link.c for machines that are missing them. Was defineing the wrong thing. Ripped out the prevention of term sending five zeros. I had made a mistake. Term only interprets the five zero's if they are line noise. Instead, add a 'breakout' option to termrc. This sets the character to watch for. Fixed the sig_child handleing a bit more. Shells should exit cleanly now... Improved the sigwinch handleing for sun's. Sun's are broken. I am getting a VERY low opinion of suns.... 099l Fixed a minor bug in link.c that caused upload to hang. It only showed up when ther was a '@' in a byte, on a 2047 byte boundry. 099m Changed the sample TERMRC to be a real sample. Fixed tmon to not core-dump is there is no compression. Changed checkline to start from 0 instead of 32 (to fit in with removeing the automatic escapeing of 0-31) Jeff Grills improved upload some more. Added a progress indicator. Now to write download....... Default timeout is now 3.5 seconds instead of 2.5 (i.e. timeout 70) This is to make it work by default with slower modems. Added lots of support for people with multiple term's on the same machine. Now if you run 'term home' it will try and read ~/.term/termrc and then ~/.term/termrc.home. It is ok for one or both of them not to exist. If the TERMSERVER environment variable is set, then clients will act as if they had been invoked with '-t $TERMSERVER' and try to connect to that server. i.e. if you run 'term home', then doing "export TERMSERVER=home" will cause all clients to use the 'home' server. Added the 'chdir' option to termrc. 'chdir ' will cause term to try and chdir to that directory on startup. This will cause all relative file references to be relative to that directory. Fixed PROTOCOL.unix. Removed the C_EXEC entry as it is obselete. Had a go at improveing the robustness of checkline/linerem. Made them work asymetrically. Improved the pty handleing. It now checks to make sure both the master and slave can be opened before commiting itself. This lets it handle ptys that are left in a dud state. Fixed a bug in serial.c. The byte_shift was resulting in corrupted bytes in seven_bit mode if byte_shift was > 127. Implemented byte stuffing. In seven_bit mode, term now stuffs 7 eight bit bytes into 8 seven bit bytes. It doesn't check if the stuffing is needed tho. :( It should only stuff if the top bit is being used, but it stuffs all raw characters. Ppl can live with this for now. Added a CPS to the upload -vv display. Fixed a minor bug in upload.c relating to -vv display. (CPS was -ve and too high ). Added a few patches to support AIX. Minor stuff. Fixed a bug with responce which resulted in results being sent to the wrong end. 099n Updated the CREDITS file. Re-wrote the intro blurb. Added sevenbit.o to Makefile.{sun,sgi}. An oversight. Fixed a litte bit of AIX support. Fixed the sevenbit byte stuffing. upload has a few bug fixes. The CPS bug is fixed. There is a '-u' flag. This will erase the file after a succesful upload. source file pathnames are now truncated. use '-as' to override this. Note that you need to use a '--' before the first filename if you want to use -as. i.e. 'tupload -f -r -vv -- local.file -as remote.file'. Extensively revised the man pages. Checkline/linerem got re-written by Jeff Grills. Try them and see what they say about your line. 099o checkline/linerem replaced by linecheck. MANY thanks to Jeff Grills. Got rid of the 'Clearing Dict table' messages. Fixed a bug relating to connections getting closed before the buffers were flushed. Deleted the checkline/linerem programs and associated man pages. Fixed a few minor bits in linecheck. Stuffed up all the formatting to suit me and my editor. Placed term under Gnu Copyleft. Changed README to suit. Added a copy of the GNU copyleft. Removed some more debugging info from main.c the 'client gave this:..' and 'select: ..' warnings are now gone. Make tredir/txconn ignore SIGPIPE signals. These were causing it to exit when clients exited in a hurry without letting tredir know about it. Changed redir, and xconn to have the capability to redir multiple ports in the one redir. If anyone wants to play with writting a user interface, feel free. :). do_connect() now takes an array of ports to watch. Re-wrote the checkline/linerem section of the README to be about linecheck. Edited the info jeff sent to be more readable for novices. 099p There was a race condition that occured when a remote client, and a local client started up at the same time. To fix, edit your ~/.term/termrc file on ONE end, and add the line 'remote' to it. This should ensure lines like 'trsh -- tupload -f -c -v o' will work. Makefile fixed to generate 'linecheck' not 'check'. More debugging info got lost. This time the perror("accept"); Installed a new(er) version of linecheck. Thanks again to Jeff. Fixed the dependancies in the Makefile for linecheck. 099q Fixed a bug in upload. It wasn't closeing files after it had uploaded them. Added -l option. Does the same as redirecting stderr. Just for ppl with braindead shells. Made term use only stdin, instead of stdin/out. I.e. term will try to write to stdin to write to the modem. This should normally be ok. Added the -1 option. If used, term will use stdout as the modem, instead of stdin. Added -v option. Usage is '-v'. Set the modem device. i.e. run as 'term -v/dev/ttys1' to use /dev/ttys1 as the modem. Warning!!! For forward compatability do NOT use more than one -v. Properly fixed upload, and added a C_CLCLOSE command for clients. Interested ppl can see link.c. Fixed a bug in send_command() that was resulting in dud command_result returns. Re-wrote the error reporting in link.c. Replace do_return with the two routines ret_ok(), and ret_fail(). term 1.0.0. YES!!!! at last! Patched in Jeff's linecheck patches. Fixed 2 bugs relating to upload. released 1.0.0 NOTE!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! ALL FUTURE RELEASED WILL BE NAMED "SLAP" There will not be another version of term. It has been re-named "slap" (for serial line access protocol). term 1.0.1 Re-wrote the stdin/out handleing as it was causeing problems on the NeXT. Fixed a small bug in upload relating to uploading a file from stdin. Note! The above implies there will be no work done on term. This is just to say that bugs will be fixed as they appear, but term will get no new features. All my efforts will now go into Slap. 1.0.2 Fixed a typo in terminal.c Had in mypid; instead of int. Thanks to warwick for this. Added code to define STDIN_FILENO in linecheck.c if it wasn't already defined. Fixed the Makefile.sun for linecheck. Oversight. 1.0.3 Fixed the problems with term hangin on Sun's. Turns out it is a dogey sun select(), and a missing continue in my source. Thanks VERY much to one David E Wexelblat. 1.0.4 Changed the way term worked for non-vhangup cases. Merged in Hugh Secker-Walker's patchs to support NeXT boxs. Merged in support for SGI machines. Added makefiles for mips, and next machines. Thanks again to Hugh. 1.0.5 Added SVR4 support. Tried to fix up a bug in xconn and tredir 1.0.5.a [ by Hugh Secker-Walker, for NeXTstep, not part of Michael's release ] At Michael's suggestion to fix a bug with high-speed links, changed link.c line 94 from 'if (p_in[.... ' to 'while (p_in[....'. Added README.NeXT, README.MIPS, select.c.mips, upload.c.mips. 1.0.5.b [ by Felix A. Lugo, E-mail: Felix_A_Lugo@ATT.COM ] Added Solaris 2.1 support. Fixed problems with do_resize (misc.c) and added support for pixel window sizing (misc.c, link.c, trshell.c). Fixed problem with restoring tty settings after abnormal exits and re-enabled SIGINT and SIGQUIT. 1.0.6 Merged in changes as per above, and released as 1.0.6 1.0.6.a [ by Hugh Secker-Walker, hugh@mit.edu ] Trsh accepts -s option for a simple connection, i.e. no pty on remote end using C_EXEC. Also, trsh restores terminal when killed. In link.c, support for C_EXEC, to run a command using /bin/sh -c, and no pty. This for the -s option of trsh. In pty.c, fixed use of pipe used to confirm that child exec'ed. Unresolved problems remain (on NeXT) with printfs in child hanging. (This latter was not a problem until the pipe bug was fixed so that the parent (term) really waits.) Also, added open_socket() for support of C_EXEC. In term.1, term_clients.1, and term_setup.1, numerous editorial changes and additions. Merged in changes as per above, and released as 1.0.6 1.0.7 Fixed Makefile.sun to look for linecheck.c instead of checkline.c Fixed misc.c to NOT put out literal bell's (dunno about you but I hate the noise). (all above due to jeff grills.) See above list of change by Hugh as 1.0.6a