From: Johnie Stafford To: All Msg #92, May-29-94 12:04AM Subject: Msdos-jnos Lives On James Dugal (N5KNX) announced on friday that he is picking up where Johan left off with the DOS version of JNOS. The following is his release announcement. I have placed the sources on 1:3803/7 (Circle of Support, New Iberia, LA (318) 367-9916). I am working on getting the update version available there as well. I have to get the BBS to accept a zip file which is a subset of another zip file. The BBS keeps seeing this as a duplicate of the complete source. So look for it in the next day or so. I am also willing to put together a couple of configurations of compiled versions if I get a chance. 73 Johnie kb5yvf@k5arh.#lftla.la.usa.noam kb5yvf%w5ddl.aara.org@usl.edu ======================================================================== While Johan is honeymooning and moving to Japan, the inexorable march of progress continues! I've taken on the challenge of maintaining the MSDOS version of JNOS. I don't plan to add a whole lot of new features, but I do want to fix bugs in what we have now. To this end, I am releasing JNOS 1.10d in both update (upd110d.zip) and complete (jnos110d.zip) form. I have placed these files on PC.USL.EDU:/pub/ham/jnos110d, FTP.ECE.ORST.EDU:/pub/ham/wg7j/incoming UCSD.EDU:/hamradio/packet/tcpip/incoming Johan asked me to change the readme.now file to request bug reports be sent to my address, shown below, rather than to his (defunct) email address. 73, -- James Dugal, N5KNX, jpd@usl.edu or n5knx@k5arh.#lft.la.usa.noam 25 May 1994 ------------------------------------------------------------------------ What follows summarizes the changes in JNOS 1.10d : forward.c ignore all connect-script lines when scanning for area names in dorevfwd(). Log exits after fwding. More tests added to check if forwarding needed. mboxmail.c handle null areaname case in changearea(). fix invalid free(rhdr). Add MC/MM to regular file. mailbox.c assign m-type as TELNET_LINK instead of TELNET, to fix MORE prompting. Log exits from mbox. mailbox.h added lockcnt to mailbox structure to support bmutil.c locking changes. bmutil.c improved locking of areas. Char-at-a-time mode when reading the telnet more prompt response. Remember position and msg delete status when new mail arrives. sockuser.c keywait() fixed to look at RX queue when flushing. alloc.c morecore() error msg fixed. Eliminate excessive logging. Applied G8FSL's MULTITASK patch. pc.c handle TABS when statuslines are used. version.c now version 1.10d. --------------------------------------------------------------------- Explications: I found that dorevfwd() in forward.c would not reject all the connect script lines in FORWARD.BBS when it was looking for area names. Hence it would try to lock areas like "+Connected to", etc. Fixed. Also I added a log entry when the mbox is exited after forwarding, to assist log analysis. changearea() in mboxmail.c assumed that the current area was valid, and invoked closenotes() on it before switching to the new area. This is not true for the first call within the dorevfwd() loop, for example. So I modified changearea() to first test for a non-null area name. While I was at it, I added the ability to move/copy to a regular file (ie, non-area) msgs using MM/MC. As always, you must have sysop privs. Fixed an invalid free of rhdr in mboxmail.c dosend(), seen when an SR command would time out. I was wondering why the more prompt was not behaving like it used to, for telnet and tip accesses to the mailbox, and I found that in mboxmail.c the code tested for TELNET and other times tested for TELNET_LINK. I changed all references to TELNET_LINK and TIP_LINK. I made a similar change in mailbox.c, and now the more prompt for telnet users again permits a single space to signal it to send the next page, and a q allows an exit. I've changed how and when a mailbox/area is locked, in bmutil.c, to eliminate calling SyncIndex without a lock being held. I was trying to eliminate possible ways for the index to not match the mailbox itself. I am not sure this is the "final cure" but it IS needed. I added lockcnt to mailbox.h, thus all other modules referencing mailbox.h must be recompiled. Fixed readnotes() to remember position in area, and which msgs were deleted, when new msgs arrive. Changed keywait() in sockuser.c to test the input queue length rather than the output queue length, when flushing input. Changed the "Mail file busy, abort or retry?" code in bmutil.c to instead elicit a y/n response, by using the telnet or ax.25 keywait routines. The query is now: "Mail file busy. Retry? (y/n)". morecore() in alloc.c was not promoting an unsigned int to an unsigned long, leading to humongous values in the display of bytes requested and coreleft. I fixed the computation. Also, I've seen 2MB log files full of morecore()'s complaints, so I changed it to log only the first unsuccessful morecore() (provided Memdebug is ON). Andrew G8FSL noted that the free(mallocw(Minheap)) technique to make heap available to Jnos while multitasking, was broken by the new free() code, and supplied a fix. I found that when status lines are in use, TABs were being displayed as a diamond symbol, instead of moving to a tabstop. I changed pc.c display() to do this properly. ANSI sequences however are still not handled due to the direct screen writes done in display(). Also I left in the debugging log() calls in bmutil.c to diagnose why X-Forwarded-To headers are sometimes not added to msgs just forwarded (which leads to redundant forward attempts later on). Compile with -DDEBUG_HDR to enable. ... A feature is a bug with seniority. ___ Blue Wave/QWK v2.12 --- QScan v1.065b * Origin: Circle of Support * New Iberia, La. (318)367-9916 (1:3803/7) More [Y,n,=]? .