946.6.1
(ATE) Fix timing issue causing ATE to sometimes close on startup before initiating the connection.
946.6.2 ATE
(ATE) Support ZTERM enable/disable serial pass-through feature:
? CHR(27);CHR(27);chr(3);PARAMS$;chr(0); ! enable passthru
? CHR(27);CHR(27);chr(4); ! disable passthru
PARAMS$ is of the form:
"<port#>,<baud>,<parity>,<databits>,<stopbits>"
For example:
"1,9600,E,7,2" (COM1, 9600 baud, even parity, 7 bits, 2 stopbits)
When enabled, input from the serial port is merged into the keyboard channel input. Output characters are not normally sent to the serial port. To enable that, send ^X (ASCII 24); send ^T (ASCII 20) to disable. Note that these are the standard AUX PORT ON/OFF commands for the AM6x emulation, but when they are sent while serial passthru is enabled, they have the effect of sending the output to the defined serial port rather than to the printer port.
946.5.1 ATE
(ATE) Fix problem with intermittent dropped keyboard character (usually at the start of an INFLD operation). (Does not require updating server, only ashw32.exe.)
946.4.1 ATE
(ATE) If host-initiated FTP transfer fails due to login parameters, it now pops up a dialog allowing you to enter them.
946.4.2 ATE
(ATE) Fix problem with ATE client using up 100% if the CPU time; fix problem with ATECTL/S not working reliably.
Both client and server sides should be updated to 946.4 at same time.
946.3.1 ATE
(Windows/ATE) Clean up various glitches in the new ATE internal protocol. Reinstate reverse channel support. Fix a bug with large XTREE responses (as with multi-select and editable cell.)
946.3.2 ATE
(Windows/ATE) Y/N and 1/0 were not working with INFLD checkboxes.
946.3.3 ATE
(Windows/ATE) TAB(-1,160) is now supported by ATE (to set cursor shape). However, it is preferable to use the new TAB(-10,50);code;chr(127); command, as it is follows the standard ATE extended command syntax and protocol. The code is a single byte, whose ASCII values can be one of the following:
32 (space) white arrow
40 "(" hourglass
48 "0" I-beam
For example, while performing an operation that may take a few seconds, like building a list of customers to load into an XTREE, you can set the cursor to an hourglass by:
PRINT TAB(-10,50);chr(40);chr(127); ! set cursor to hourglass
946.3.4 ATE
(Windows/ATE) The abbreviated TAB(-10,20);string;",DL";chr(127); format now strips trailing blanks (except for 1) before right justifying.
946.1.1 ATE
(Windows/ATE) A new (and improved!) internal protocol for communication between the application and GUI front-end has been implemented, which may eliminate the motivation for the reverse channel. Indeed, the reverse channel has been disable while this is being studied. (ATECTL/C will return a failure code immediately, so the application will just revert to using the standard keyboard channel.)
Briefly, the new protocol tries to address the problem of typeahead getting mixed in with GUI command responses by putting the client side into a keyboard "hold" at the start of a sequence of GUI operations and leaving it there until the server side of the application requests actual input (e.g. INPUT LINE, XCALL ACCEPT, XCALL INFLD, etc.) During the "hold", local keyboard operations are handled normally but just held on the client side (not transferred to the server side of the application).
The only known difficulty with this scheme is that it exposes a more likely possibility that some unexpected sequence of events will leave the application waiting for input while the keyboard hold is still active. This appears to the user as if the keyboard is just locked, and can be manually fixed by clicking on the Settings|Keyboard Locked menu toggle. (The hold shares the "keyboard locked" menu toggle with a true keyboard lock, but unlike a true keyboard lock, in this case keyboard characters are still accepted and buffered so that typeahead and can be forwarded to the application in a coherent manner when the appropriate time comes.)
Note that although this is mainly an issue for ATE, where the GUI client and application side are on separate computers, the paradigm still applies even in the standalone Windows case whenever you use TAB(-10,x) commands (and sometimes you may not realize you're using them because they are internal to some XCALLs and LITs).
946.1.2 ATE
(Windows/ATE) Related to the new keyboard buffering system, there is a new MIAMEX call:
xcall MIAMEX,153 {,op}
If op is zero or omitted, this releases the keyboard hold, otherwise it sets it. In theory you should never need this command, since A-Shell should take care of it automatically, but it might be useful in some exotic situation.
946.1.3 INFLD
A minor bug in the non-GUI INFLD SETDEF display, in which the list of choices started with some apparent garbage characters, has been fixed.
946.1.4
(Windows) Fix a printing problem introduced in 944 in which specifying copies > 1 and using the printer selection dialog could lead to the actual number of copies being the square of the requested number. (This was caused by both A-Shell and the printer driver both implementing the copies.)
946.1.5 ATE
(Windows/ATE) The right justification logic for static text controls has been refined to avoid the appearance of the right-justified text being jammed too closely to the end of the control. (This was really just an aesthetic issue with the MBF_SUNKEN style.)
946.1.6
(Windows) When a SUBMIT'ed job runs into a: K (command file keyboard wait) it no longer aborts the process, and instead just ignores it. It makes no sense for a background job to wait to stop and wait for keyboard input, so this is mainly a convenience for controls files that indirectly execute START.CMD or other command files that use: K to allow the operator to see a message before continuing. (This actually restores the way it worked under 4.8).
Note that under UNIX, SUBMIT'ed jobs use the redirected standard input in place of the keyboard, so a: K will pick up the next line from the standard input.
946.1.7
(Windows/ATE) Fix a problem introduced in 945 in which radio buttons would change into checkboxes when selected within EVENTWAIT.
