TSO Kermit 4.3.x "Beware File" - Known Bugs, Problems, Limitations. As of: 13:35 Wednesday, 2000 Sep 27 This file contains (A) instructions for applying Kermit updates, (B) a list of unresolved problems, (C) a list of problems with work-arounds or optional fixes, and (D) a list of resolved problems. NOTE: entries will sometimes be entered under (D) even though the needed updates have not been released in IKTKER.UPD. Please check the date of the UPD and NUP files to see if they have been changed before requesting new copies. Bug reports may be sent to John Chandler . Kermit-370 carries an "edit" number in addition to the version and release numbers. The three numbers together are printed out in the assembly listing and displayed by Kermit at startup time. Be sure to include all three numbers (e.g., 4.3.0) along with any error reports. A) Update instructions ................................................. Many updates have been received since Kermit-TSO 4.2 was released. All updates prior to 93/9/30 have been removed from IKTKER.UPD and applied to the sources in the process of creating 4.3. If your source does not include the line defining &KVRSN to be 4.3, you should obtain an all-new source distribution from Columbia. Any updates created later than your source should be applied before you make the executable module. In general, updates received at Columbia will not be applied to the source files in the TSO Kermit distribution until and unless "major" bugs are uncovered, and then all accumulated updates will be applied at once. The description of each problem and a pointer to its solution will always be included in IKTKER.BWR (the present file), but the updates themselves are to be found in IKTKER.UPD. The procedure to update your copy of TSO Kermit is as follows: 1) Obtain the latest version of IKTKER.UPD, which contains all the updates as a composite file punctuated by comment lines introducing each update's name and briefly describing its purpose. Please note that some modes of file transport have a side effect of either doubling or removing any periods in the first column -- if your copy of IKTKER.UPD has records beginning with "/", then all instances of "/" in column 1 should be changed to "./", and all records with "@" in column 63 (introducing an update code normally found starting at column 64) should be moved over to make room for a "." in column 1. Similarly, any instances of ".." in columns 1-2 should be changed to just ".". A related problem can affect the sources by leaving a blank in column 80. That can happen via the loss of a "." in column 1 (in which case the sequence number will be shifted into columns 72-79) or via automated efforts to preserve the "." (in which case the sequence number will be truncated). The repair consists of inserting the missing "." in column 1 or "0" in column 80. 2) Determine the date of your sources. If you obtained the source from Columbia, the date is the value of the &KDATE variable symbol in the source itself, i.e., the release date of the current base version. If you received the source distribution from a secondary supplier, you will need to examine the code to determine the most recent update applied. Just sort the combined source by columns 67-71; the last line will be labeled with the last-applied update. 3) If your sources are "pure" Columbia 4.3, go on to step 4. Otherwise, you must take care to avoid applying updates that are already in the source. If it is inconvenient to obtain "pure" sources, you must go through the list of updates at the beginning of IKTKER.UPD and remove the entries for updates already included in (i.e., older than) your sources. Each line of the list begins with "./ *" plus an update code in the form xxyyddd (the yyddd is just the date of the update -- the list is arranged chronologically). Also, go through and discard those updates from the file. Each update begins with a second copy of the line from the list and goes to the next such line. The updates are also in chronological order. 4) Extract any desired optional updates from IKTKER.BWR (this file) and group them together with any local modifications specific to your site. NOTE: all updates in this file are indented one column to avoid all the transmission problems mentioned above in step 1. Be sure to shift everything over when you extract the updates. 5) Apply the updates to your composite source by running GUPI. There is a batch job in IKTKER.INS for installing Kermit from scratch, and the same job can be rerun as new updates become available. Simply omit job steps CNVASM through GUPLNK in subsequent runs. Note: Kermit uses the STATUS macro, which is sometimes found in SYS1.AMODGEN, instead of SYS1.MACLIB. Further note: be sure the macro libraries you use are compatible with the assembler -- at some installations, there is apparently a mismatch, such that the GETLINE macro has a definition with an "invalid" parameter name. Also, the GETLINE and STAX macros supplied with MVS/XA have been "modernized" to the extent that they cannot be used with the VS assembler. Further still, the GTSIZE macro is mangled at some sites and may require APAR OY22935. If you prefer, GUPI can be run interactively as a command processor. In that case, the updated source would still have to be assembled. In any case, you should not discard the original source because the updating process is simplest if you always start from the same base source. NOTE: step 5 requires that you have version 1.1 or later of GUPI, which in turn means that your sources must be no older than 88/9/15. ------------------------------------------------------------------------ B) Unresolved problems ................................................. ---------------- 88/3/9 ------------------------------------------------ An ABEND within a TSO command called from Kermit propagates back to the TMP, which then displays a READY message. Typing a carriage return will resume Kermit operation. ---------------- 89/1/20 ----------------------------------------------- Now that Kermit-370 supports multiple-filespec SEND, it is possible for a single SEND subcommand to specify overlapping wildcard filespecs. Only the CMS variant keeps track and makes sure that it doesn't send the same file twice within one send invocation. The others, including TSO Kermit-370, may send the same file again, and it is left to the user to prevent filename-collision problems on the receiving end. ---------------- 90/5/8 ------------------------------------------------ With a few exceptions, CONTROLLER will be set automatically when Kermit is invoked. See file IK0AAA.HLP in the Kermit distribution for a list of front ends that have been reported to allow Kermit file transfers and another list of reported failures. In those lists, CONTROLLER type VTAMTTY is omitted -- that being a peculiarity of TSO Kermit and the same as TTY, but needed when connected through TSO/VTAM if it does *not* have the IKTWTX1 exit. ---------------- 91/8/16 ----------------------------------------------- The type-out interception mechanism for server mode works only for "normal" line-oriented output, not for fullscreen applications. Indeed, since there is no mechanism provided for feeding subcommands to a program on the mainframe through the Kermit server, only one-shot, non- interactive commands should be issued via REMOTE HOST. ------------------------------------------------------------------------ C) Problems with workarounds or optional fixes ......................... If it becomes necessary to diagnose problems by turning DEBUG on, there may be trouble in handling the resulting KER.LOG dataset with certain tools, since the log is allocated with an LRECL equal to the largest allowed packet size. In practice, though, diagnostics can usually be done with short packets, so that the dataset can be copied with a truncated LRECL (e.g., 84). Even if long packets are present, it may be possible to truncate the records without losing the important details. ---------------- 88/3/9 ------------------------------------------------ The REMOTE COPY function of a Kermit-TSO server and the TYPE subcommand both require the TSO data utilities package available from IBM and other vendors. The necessary commands are COPY and LIST, respectively. If your system doesn't have them, it should be possible to create your own. For example, you can write CLIST's to reside in your CLIST library. ---------------- 88/3/9 ------------------------------------------------ While CLISTs may be invoked under Kermit, if they are called from within a TAKE file, they are not then executed until Kermit returns to command level. This means that a CLIST invoked from the system KERMINI file would execute after the user KERMINI. If such a CLIST is necessary, it would be advisable to comment out the first BAL 9,LUPTIN in USNTRF (to suppress the direct execution of the user KERMINI) and to add code to the end of the CLIST to execute the user KERMINI (if any). ---------------- 88/7/18 ----------------------------------------------- On some systems, TTY lines connected via VTAM are unable to carry long packets outbound because of the insertion of CRLF's every 130 bytes. When that is the case, it may be possible to circumvent the problem by applying the following update to the Kermit source (although the IBM documentation states that setting the line size to a value greater than 255 has unpredictable consequences). ./ * SC88200 - Correct line length problem ./ R 05728000 $ 5728000 1 ->4.3<- STMSTY STSIZE SIZE=1920 Set up linesize @SC88200 ---------------- 89/2/27 ----------------------------------------------- The code for detecting the difference between SERIES1- and GRAPHICS-type terminal controllers may cause peculiar error messages at session startup, particularly if you wish to run Kermit on an IBM 3270-type terminal. Normally, Kermit can recover from these error conditions, but some configurations and/or versions of VTAM may hang the terminal session instead. For example, there may be an error message IKT00405I SCREEN ERASURE CAUSED BY ERROR RECOVERY from VTAM that never goes away. In that case, there are several possible cures. First is the simple expedient of pressing PA1 to clear the error condition. If that doesn't help at your site, you may find that logging onto the host via Netview's TAF component makes the error message appear only once. If you cannot persuade VTAM to be more sensible, you might want to bypass the code (especially if you don't have any GRAPHICS-type terminals). The code is governed by a set of switches which can be tailored to your needs: STCQBIT EQU X'01' WSF Query is always allowed STCQNS1 EQU X'02' WSF Q implies *not* SERIES1 STCS1 EQU X'04' Always assume SERIES1 STCNOS1 EQU X'08' Always assume *not* SERIES1 STCNORD EQU X'10' Asynch READ MOD is forbidden The standard flag value on TSO is STCQNS1; if you wish to force the code to select SERIES1, you should change that to STCS1, as in the following: ./ * SC9131X - Customize controller type selection ./ R 01062000 $ 1062000 100 11/21/91 ->4.3<- &CONOPTS SETC 'STCS1' SETCON options @SC9131X Alternatively, if you have *only* GRAPHICS-type controllers, you could change it to STCNOS1. ---------------- 89/5/19 ----------------------------------------------- A mode of operation favored by some ISPF users is to keep files in one PDS, rather than scattered among a variety of PDS's and sequential files. In that mode of operation, Kermit TAKE files may not have a file extension of TAKE or KERMINI, but the user KERMINI file, in particular, must have a proper file extension in order to be recognized by Kermit at startup. For example, if each user had a PDS called ISPF.TEXT, and the user's KERMINI were expected to be member KERMINI of that PDS, the following update would make Kermit-TSO look in the right place (and also allow other TAKE files to be kept in ISPF.TEXT). ./ * SC89139 - Alternate location for init file ./ R 01020000 $ 1020000 200 05/19/89 ->4.3<- USRTAKE DC C'ISPF.TEXT(KERMINI)' Init file @SC89139 ./ I 05141000 $ 5141010 10 05/19/89 ->4.3<- CLC =C'.TEXT',3(1) Or is is .TEXT? @SC89139 BE RTRN0 That's ok too @SC89139 ---------------- 89/9/27 ----------------------------------------------- Kermit-TSO will recover data sets migrated using DFHSM, but not those archived by the ABR software. As a result, if the KERMINI files are no longer on-line, Kermit will simply miss them. The following update will cause data sets to be recovered as needed, provided Kermit has not already entered protocol mode. ./ * SC89270 - Recover ABR-migrated data sets. ./ D 06023500 06024000 09/27/89 ->4.3<- ---------------- 90/3/1 ------------------------------------------------ On some systems, the leading DSN qualifier SYS1 is restricted, so that the system-wide Kermit INIT file cannot be cataloged as expected by Kermit-TSO. In such cases, the following update (or something similar) can be used to substitute the prefix of your choice. ./ * SC90060 - Alternate prefix for global init file ./ R 01018500 $ 1018500 100 03/01/90 ->4.3<- SYSTAKE DC C'''SYS2' @SC90060 ---------------- 90/4/15 ----------------------------------------------- Some micro Kermits may garble the date/time stamp transmitted with files downloaded from TSO because there is only a date and not a time. As a matter of fact, some files have times as well (at sites with a system update that maintains a "modification date"), but many do not. This problem has been reported for some versions of ProComm, but MS-Kermit from Columbia is OK. There are two workarounds: switch micro Kermits or issue SET ATTR DATE OFF to TSO Kermit. ---------------- 90/5/11 ----------------------------------------------- If you feel it is too dangerous to allow a remote Kermit to issue a wildcard DELETE command to TSO Kermit, you can disable that feature via the following update. The effect is to treat any asterisk in a REMOTE KERMIT DELETE command as a number sign. Beware if you ever use a bare number sign as a DSN qualifier. ./ * SC90131 - Disable wildcard delete ./ D 05128000 05128500 05/11/90 ->4.3<- ---------------- 90/7/5 ------------------------------------------------ Kermit-370 now honors the file type specification (binary or text) given by a sending Kermit in an Attribute packet. Old procedures for sending binary files may therefore not work properly anymore. To return to the old mode of operation, issue the subcommand SET ATTR TYPE OFF. To live with the new mode, be sure to issue the SET FILE TYPE to both sending and receiving Kermits (to be safe). ---------------- 90/7/30------------------------------------------------ When the file type is TEXT and the recfm is F, Kermit-370 strips off trailing blanks from each record before sending a file. In order to keep trailing blanks in a downloaded file, you can convert it to RECFM V in TSO before sending it, e.g., by using COPY with the RECFM VB option. This is true of versions from 4.1 onward. ---------------- 90/8/22 ----------------------------------------------- If, for some reason, the other Kermit insists on pretending ASCII bytes all have the 8th bit set, Kermit-370 will not detect the intended CRLF separators between lines. The result of uploading from such a Kermit is a (text) file with only one record. If the other Kermit cannot be fixed, the following update will make Kermit-370 treat ASCII 8D and 8A the same as 0D and 0A, as long as the ATOE table does so. ./ * SC90151 - Forgive 7-bit ASCII files with 8th bit on ./ I 02490500 $ 2490700 200 08/03/90 16:06:46 ISW78 DC X'7F' Start with 7-bit ASCII @SC90151 ./ I 02713500 $ 2713700 200 08/03/90 16:06:46 SW78 DS X'7F' Switch for 7-bit/8-bit ASCII @SC90151 ./ I 03347500 $ 3347700 200 08/03/90 ->4.3<- KCALL TBLCRLF Note current treatment @SC90151 ./ I 03350500 $ 3350700 200 08/03/90 ->4.3<- KCALL TBLCRLF Note current treatment @SC90151 ./ I 04016000 $ 4016200 200 08/03/90 ->4.3<- KCALL TBLCRLF Reset 7-bit/8-bit mask for CRLF @SC90151 ./ I 04063000 $ 4063040 40 08/03/90 ->4.3<- * * Entry: nothing special @SC90151 * Exit: SW78 set up correctly @SC90151 TBLCRLF ENTER ALT @SC90151 MVI SW78,255 Assume 8-bit codes @SC90151 CLC ATOE+ALF(1),ATOE+ALF+128 @SC90151 BNE RTRN0 @SC90151 CLC ATOE+CR(1),ATOE+CR+128 @SC90151 BNE RTRN0 @SC90151 MVI SW78,127 No, 7-bit codes @SC90151 B RTRN0 @SC90151 ./ I 08391000 $ 8391200 200 08/03/90 ->4.3<- NC CUR,SW78 If 7-bit ASCII, ignore 8th bit @SC90151 ---------------- 91/1/4 ------------------------------------------------ There may be problems with timing that can be avoided by removing the capability to timeout from the mainframe side as in the following optional update. This update is part of the official release of Kermit for ROSCOE and, in that environment, avoids multiprocessing abends. ./ * SC91004 - ROSCOE IDENTIFY problem and FULLSCREEN timeouts (TSO) ./ D 01027500 01032000 01/31/91 ->4.3<- ./ R 05529000 05538000 $ 5529000 1000 01/31/91 ->4.3<- L 0,4(,1) Size of TGET buffer @SC92030 L 1,0(,1) Adr of buffer to putin @NW86330 TGET (1),(0),ASIS @NW86330 LTR 15,15 @NW86330 BZ TRMRET Ok @SC91004 C 15,F12 @NW86330 BE TRMRET Ok @SC92030 CH 15,=H'24' @SC92030 BNE RTRNM1 Signal I/O error @SC91004 TRMRET ST 1,KTGETT+4 Store length read @SC92030 ./ D 05541500 05544500 01/31/91 ->4.3<- ./ D 05549000 05560500 01/31/91 ->4.3<- ./ D 05744500 05749000 01/31/91 ->4.3<- ./ D 05768500 04/11/91 ->4.3<- ---------------- 91/9/18 ----------------------------------------------- There is a report of IKT00405I errors (see optional update SC9131X) in situations that have nothing to do with unexpected controller types. For example, the screen erasure error may occur in interactive mode if a Kermit subcommand is entered on the last line of the screen. Pressing PA1 appears to clear the error condition. ---------------- 91/11/21 ---------------------------------------------- In order to change the default Kanji encoding from IBM-style to another, add one of the following updates. ./ * SC9132X - Change Kanji default ./ R 01416000 $ 1416000 10 11/21/91 ->4.3<- &KNJDEF SETC 'HITACHI' Default Kanji @SC9132X ./ * SC9132X - Change Kanji default ./ R 01416000 $ 1416000 10 11/21/91 ->4.3<- &KNJDEF SETC 'FUJITSU' Default Kanji @SC9132X ---------------- 92/4/10 ----------------------------------------------- The minimum size of strings to be subjected to run-length encoding may need to be changed to two to make the new FULLSCREEN mode work on certain protocol converters, such as the MICOM 7400, which convert three successive blanks into cursor motion commands (or even two!). The following update accomplishes that. ./ * SC92101 - Encode multiple blanks ./ I 08296500 $ 8296550 50 04/10/92 ->4.3<- CLI 0(8),ABL Is it a blank? @SC92101 BNE ENCRP1 No, use normal criterion @SC92101 C 14,F2 @SC92101 BNL ENCRP2 Always encode multiple blanks @SC92101 ENCRP1 DS 0H @SC92101 ./ I 08297500 $ 8297700 200 04/10/92 ->4.3<- ENCRP2 DS 0H @SC92101 ---------------- 92/10/18 ---------------------------------------------- When Kanji characters are first selected, the default for the single- byte character set is CP290 (hard-coded in Kermit-370). That can be changed, however: ./ * SC92252 - Make CP281 the default SBCS with KANJI ./ R 04118000 $ 4118000 100 10/18/91 ->4.3<- TBVJP DC CL(LALF)'CP281',CL(LALF)'CP500',CL(LALF)'CP290' @SC92252 ---------------- 93/1/21 ----------------------------------------------- Kermit-370 in FULLSCREEN mode can be at the mercy of clever efforts toward screen "optimization" by the protocol converter. One possible symptom is the suppression of the last outbound packet in a transfer if the screen is cleared just afterwards. This may happen after an upload or upon leaving server mode. To avoid this possibility, Kermit-370 can simply wait a few seconds to give the protocol converter less excuse for suppressing the last packet. The following update uses four seconds, but the optimum time may be more or less, depending on the system. ./ * SC93021 - Try to prevent end effect in FULLSCREEN mode ./ I 08780000 $ 8780000 80 01/22/93 ->4.3<- CLI TRMTP,C'F' @SC93021 BNE INTINICN Go if not FULLSCREEN @SC93021 LA 0,4 Wait four seconds to synchronize @SC93021 KCALL SUPFNC,9 @SC93021 INTINICN DS 0H @SC93021 ---------------- 93/4/15 ----------------------------------------------- Kermit-TSO now has to face three different linemode environments: the original non-VTAM line (now seldom seen), normal VTAM LU1 lines, and a new class of SNA 3770-type lines. Kermit automatically handles the first two types by adopting the proper terminal I/O method to get the necessary control characters out. These are reflected in the TTY and VTAMTTY settings for CONTROLLER in Kermit. Unfortunately, vanilla VTAM does not let any control characters out for the third type (except, possibly, BEL). Therefore, Kermit can work only when VTAM has been equipped with one of two exit routines, IKTRTX1 or IKTWTX1. If the latter is installed (and many sites have it), VTAMTTY is not needed at all, and TTY will serve for all three types of linemode sessions. The following update disables the automatic selection of VTAMTTY. ./ * SC93105 - Disable VTAMTTY selection: IKTWTX1 installed (TSO) ./ D 05732500 ->4.3<- Alternatively, either with or without IKTWTX1, VTAM will permit control characters through the third class of lines when the IKTRTX1 exit is installed. Apparently, the widely used version of IKTWTX1 distributed by the SAS Institute will work when simply renamed IKTRTX1. Also, the following code will serve. IKTRTX1 CSECT USING *,15 TM 9(1),X'30' CHECK TPUT OPTION BNM RTRN0 NOT CONTROL - ALLOW EDITING TM 9(1),X'20' BZ RTRN0 NOT CONTROL LA 15,4 CONTROL - NO EDITING BR 14 RTRN0 SR 15,15 BR 14 END ---------------- 93/9/20 ----------------------------------------------- When issuing the STOP command to get out of Kermit protocol mode, it may be necessary to hit the CLEAR EOL (not the CLEAR) key so that Kermit-370 will recognize the command. In general, if the word "stop" is echoed to the screen as you type it, and if there is already stuff on that same line, that previous stuff will be sent to the host when you hit ENTER. To prevent that, you must clear it without causing an interrupt, i.e., without using the CLEAR key. ---------------- 94/6/15 ----------------------------------------------- Since the introduction of Kermit-370 for TSO, it has been a true Command Processor, reading its subcommands from the stack. However, the original TSO Kermit read directly from the terminal. Any site with Kermit applications requiring that Kermit not gobble up the stack, if any, can install the following update (thanks to Axel Skough). Alternatively, such applications can stack a QUIT before invoking Kermit with command-line arguments. ./ * SS94137 - preserve stacked commands for 1.0-compatibility (TSO) ./ I 05430000 $ 5430100 100 05/17/94 12:00:00 BZ RTRN0 Do not "eat" beyond substack @SS94137 ICM 0,3,0(1) Check length @SS94137 BZ RTRN0 Respect barrier element @SS94137 ./ R 05438000 $ 5438000 100 05/17/94 12:00:00 SUBSTACK=YES,MF=(E,IOPLAREA) @SS94137 ./ R 05530000 $ 5530000 100 05/17/94 12:00:00 SUBSTACK=YES,MF=(E,IOPLAREA) @SS94137 ./ I 05744000 $ 5744100 100 05/17/94 12:00:00 LA 0,APGPB Adr of GTPB @SS94137 ST 0,IOPLIOPB-IOPL+IOPLAREA @SS94137 STACK MF=(E,IOPLAREA),PARM=STK0 Set up barrier @SS94137 ./ I 05768000 $ 5768100 100 05/17/94 12:00:00 LA 0,APGPB Adr of GTPB @SS94137 ST 0,IOPLIOPB-IOPL+IOPLAREA @SS94137 STACK MF=(E,IOPLAREA),PARM=STK1 Delete barrier @SS94137 ./ I 05778000 $ 5778100 100 05/17/94 12:00:00 STK0 STACK MF=L,BARRIER=* Set up barrier @SS94137 STK1 STACK MF=L,DELETE=BARRIER Remove barrier @SS94137 ---------------- 95/11/2 ----------------------------------------------- Kermit-TSO looks for the personal INIT file under the name KERMINI, which normally is found under the top-level qualifier of the userid. However, if you run Kermit after issuing a PROFILE NOPREFIX or set a prefix other than the userid, Kermit will look wherever the prefix indicates. Such behavior is perhaps not ideal, but it is enshrined by long tradition. Kermit initializes its own PREFIX to whatever the user profile has. Here is an optional update that should force Kermit to look only under the userid: ./ * SC95306 - Use 'userid.KERMINI' for private init file (TSO) ./ I 00994500 $ 994600 100 11/02/95 00:04:35 USRTAKE DS C'''xuserid.KERMINI''' INIT file @SC95306 LUSRT DS H @SC95306 ./ D 01020000 01020500 11/02/95 00:04:35 ./ R 01077000 $ 1077000 100 11/02/95 00:04:35 MNOTE '--- IKJCPPL, IKJECT, IKJGTPB, IKJIOPL,' @SC95306 MNOTE '--- IKJPSCB, IKJUPT,' @SC95306 ./ I 01081500 $ 1081700 200 11/02/95 00:04:35 IKJPSCB , @SC95306 ./ R 03006000 $ 3006000 100 11/02/95 00:04:35 LH 1,LUSRT Length of name @SC95306 ./ I 05709500 $ 5709550 50 11/02/95 00:04:10 L 2,CPPLPSCB Get ptr to PSCB @SC95306 USING PSCB,2 @SC95306 XR 3,3 @SC95306 IC 3,PSCBUSRL Get length of userid @SC95306 MVI USRTAKE,C'''' @SC95306 MVC USRTAKE+1(7),PSCBUSER Move userid @SC95306 LA 4,USRTAKE+1(3) Advance over userid @SC95306 MVC 0(L'DKERMINI,4),DKERMINI @SC95306 LA 3,1+L'DKERMINI(,3) @SC95306 STH 3,LUSRT Length of string @SC95306 ------------------------------------------------------------------------ D) Resolved problems ................................................... ---------------- 90/5/8 ------------------------------------------------ A commonly reported "problem" is that Kermit-TSO objects to a send packet size greater than 94, and users wonder how to turn on long-packet protocol for downloads. The answer is that the the maximum packet size in Kermit protocol is always dictated by the receiving Kermit -- ALWAYS. NOTHING YOU CAN SAY TO A SENDING KERMIT WILL EVER MAKE IT SEND PACKETS LONGER THAN THE RECEIVING KERMIT ASKS FOR. Conversely, Kermit-TSO will always use packets as long as possible unless frequent transmission errors indicate a noisy communication line. The only purpose for the send packet-size parameter in Kermit-TSO is to set the default, in case the receiving Kermit is badly crippled (in which case, long packets are entirely out of the question anyway), or to control "raw" downloads (those without Kermit protocol and parameter negotiation). ---------------- 90/6/22 ----------------------------------------------- Kermit-370 supports file transfers through the IBM 3174 AEA with level 2 microcode in configuration B. The support is restricted to terminal types defined in the 3174 to have ASCII Graphics capability. That capability is available by default only for VT241 and Tektronix 4205 terminals, but user-defined terminal types are also possible. Also, the Graphics capability is allowed only for lines defined without associated Host Addressible Printers. Kermit-370 now automatically detects the B2 AEA and sets CONTROLLER accordingly (to AEA if Graphics is allowed, to NONE if not). However, if the 3174 is owned by VTAM, and the connection is made with a logmode that forbids the Read Partition Query (such as M2SDLCNQ), Kermit cannot detect the AEA and will default to CONTROLLER GRAPHICS (and, incidentally, cannot transfer files even with CONTROLLER set to AEA by hand). M2SDLCQ is known to allow correct operation, but other logmodes have not been tested. Since the 3174 supports full 8-bit communication, it may be useful to configure the ports for 8-bit data and to set both SEND and RECEIVE PARITY to NONE in Kermit-370. The fix: SC90173. Note: B2 microcode became available from IBM in May 1990. B3 followed soon afterward, and it supports file transfer the same way as B2. ---------------- 94/3/15 ----------------------------------------------- Kermit-370 now sends an explicit NAK packet when it receives a bad data packet, instead of resending the previous ACK (which is nominally the same thing in Kermit protocol). This way, there are no needless timeouts when dealing with a zealous sliding-windows Kermit. The fix: SC94074 in IKTKER.UPD. ---------------- 94/6/23 ----------------------------------------------- Kermit-370 honors a new protocol feature known as "WHATAMI". Thus, when Kermit-370 is in server mode, the user can issue a SET FILE TYPE to the local Kermit and have it take effect in the server as well. The local Kermit obviously must also support WHATAMI. The fix: SC94174 in IKTKER.UPD. See also the note in section (C) dated 90/7/5. ---------------- 94/6/30 ----------------------------------------------- Kermit-370 honors the new level-1 restart/recovery mode. When a similarly equipped Kermit is at the other end, an interrupted binary transfer can be resumed where it left off by issuing a RESEND instead of a SEND (same syntax). Attributes must be enabled at both ends of the transfer, and the sender must be set to binary (V-BINARY and D-BINARY are also acceptable). If the file in question is already fully received at the other end, RESEND does nothing; if the file is not even partially received, RESEND behaves just like SEND. In order to insure that the partial file is not erased when a transfer aborts, it is helpful to issue a SET INCOMPLETE KEEP to the receiving Kermit. Kermit-370 now also supports the UPDATE option of FILE COLLISION. With that option, a received file is rejected (on grounds of "date") if the transmitted time stamp is no later than that of an existing file of the same name. The fix: SC94181 in IKTKER.UPD. ---------------- 94/9/3 ------------------------------------------------ Kermit-370 gives more accurate transfer statistics after uploads in server mode. Formerly, the time used in figuring transfer rates included any waiting time after the transfer completed but before the next inter-Kermit communication (result: understimated efficiency). The fix: SC94245 in IKTKER.UPD. ---------------- 95/1/23 ----------------------------------------------- Kermit-370 misbehaved when doing a RESEND of a file that was already entirely received on the target system. If the length matched exactly that of the file to be sent, the whole thing would be sent again and appended. That is now corrected. Also, Kermit-370 now has the APC subcommand for sending Kermit commands to the micro while in connect mode. The fix: SC95023 in IKTKER.UPD. ---------------- 95/2/1 ------------------------------------------------ Kermit-370 objected to receiving a RESEND of a file too large to fit on disk without removing the portion already received. It now takes into account the fact that that portion will not be duplicated. The fix: SC95032 in IKTKER.UPD. ---------------- 95/2/2 ------------------------------------------------ Kermit-370 is now at level 4.3.1. ---------------- 95/2/28 ----------------------------------------------- Kermit-370 now uses the newly-transmitted time tag for a file that is resumed after an abortive transfer. The fix: SC95059 in IKTKER.UPD. ---------------- 95/4/18 ----------------------------------------------- Kermit-370 no longer treats an error-free file transfer as an "error" for the purpose of deciding whether to halt execution of a TAKE file. The fix: SC95108 in IKTKER.UPD. ---------------- 95/6/23 ----------------------------------------------- Kermit-370 now supports the REGET subcommand, which works like GET, but makes the server do a RESEND of the requested file. The fix: SC95174 in IKTKER.UPD. ---------------- 95/10/10 ---------------------------------------------- Kermit-TSO has been unable to set the working "directory" to a PDS ever since support was added for GDG's. The fix: SC95283 in IKTKER.UPD. ---------------- 96/5/1 ------------------------------------------------ Kermit-370 now suppresses the "doubled" transmission of a file that can sometimes occur on a flaky connection. (This was a problem introduced in the process of avoiding unnecessary time-outs on flaky lines.) The fix: SC96122 in IKTKER.UPD. ---------------- 96/5/30 ----------------------------------------------- Kermit-370 now announces its system type to the other Kermit before a transfer. This meshes with new versions of the various micro Kermits that expect that information and will display it to the transfer screen. The fix: SC96151 in IKTKER.UPD. ---------------- 96/6/6 ------------------------------------------------ Kermit-370 has been modified to take advantage of another recent upgrade to micro Kermits -- it will be possible to issue a SEND or GET to the mainframe Kermit, and the micro will automatically pop from terminal emulation into transfer mode and then revert to emulation as soon as the transfer is done. The fix: SC96158 in IKTKER.UPD. ---------------- 96/6/13 ----------------------------------------------- Kermit-TSO has been updated to comply with the conventions for time tags beyond 1999. Kermit-TSO still makes no attempt to set the creation date of a received data set to match the date received along with the file. The fix: SC96165 in IKTKER.UPD. ---------------- 97/1/28 ----------------------------------------------- Kermit-370 was failing to clear the "Kermit command error" status upon execution of the VERSION and ECHO subcommands. It now clears that status condition to "No errors" after all subcommands, as described in the documentation. The fix: SC97028 in IKTKER.UPD. ---------------- 97/6/13 ----------------------------------------------- Kermit-370 is now at level 4.3.2. ---------------- 97/7/23 ----------------------------------------------- The installation batch job promises to work regardless of the DCB parameters of the source files, but that was not quite true. Two files had to be FB/80 to assemble properly. The batch job has now been changed to remove that restriction. ---------------- 99/2/12 ----------------------------------------------- When downloading some PDS members, such as load modules and certain other non-ISPF files, Kermit-TSO is now more insistent on transmitting the PDS MOD date, if any, instead of the creation date, as the time tag. The fix: SC99043 in IKTKER.UPD. Kermit-TSO is now at level 4.3.3. ---------------- 99/3/11 ----------------------------------------------- TSO sessions would sometimes hang while doing a Kermit transfer. The problem lies in VTAM 4.4.1. The fix: PTF UW49973 from IBM. ------------------------------------------------------------------------