Rexxmous - Rexx Mouse DLL

Rexx Mouse DLL - Basic Text Window Mouse Interface - 11-18-2009 by DGD Binaries and Virtual Pascal Source - Freeware for OS/2 and ECS systems only. I've discovered several times that mouse support for plain text Rexx is so near zero as to be zero. There's only one package on Hobbes that has any, and it only returns which _line_ was clicked on. I suppose the notion falls into the gulf between those who can use the command line and... the rest. Anyway, _basic_ mousing is desperately needed for a (real, VIO, monospaced) word processor that I'm working on, and so, in desperation, found instructions on the web on building Rexx extension DLLs, and downloaded a shiny new copy of Watcom 1.8 to compile the code. Unfortunately, the instructions were for the Visual Age compiler, and I couldn't figure out the linking stage (though that may change, intend to look at again), so went back to Virtual Pascal that DOES have a working Rexx DLL example, I've just been ignoring it like everyone else has. Looked like all I needed to do was the calls to get basic mouse data. A few obstacles presented themselves. First, like all modern compilers, you can't JUST access the mouse through the interrupt services. Well, you CAN, but re-inventing that wheel, even with TP5 code handy, was about as difficult as using a provided routine, once I had dug it out of the heap of units and functions (there are at least THREE levels for getting mouse info). I chose the mid-level because found it before the lowest, and the high level requires more complications from a vast library. Soon had a working ordinary program to test (included here), then learned what is meant by "MouReadEventQue": it returns a STREAM of events reporting mere mouse motion, whether you want them or not, and I didn't. After consideration, decided that would eventually be polling and have to toss non-clicks anyway, so what was needed was to do most of the tossing at compiled speed rather than in relatively slow Rexx code. Came up with a simple way to discard all unwanted events. Fine. Arrange the data into strings, ready to convert into DLL form. ... I omit MUCH trial and error, most of it useless because Rexx DLLs are locked into memory and prevent further testing unless extensive re-naming is done. -- Eventually got to a version that at least ran, but shortly after, not only lost the exact incantation but so confused the system that I couldn't use my ordinary Rexx command line tools that rely on RexxUtil. It's not pleasant to re-boot this system: have several torrents running that should be manually stopped, and a balky but not flaky Seagate 250G drive adds nearly four minutes. Because of the long painful re-boot cycle, spent more than usual time analyzing the problem and looking for another option. Appears to be no way to get Rexx to release a DLL except to re-boot (except maybe RexxUtil, but don't care, not the problem). Then WASTED time because "prettied up" code toward final form, unwittingly causing more trouble than all previous to even get back to working at all. Rexx DLLs are a quite murky area, but I can now authoritatively state two crucial points: DON'T use "SysLoadFuncs" for the name of initialization function as the VP example code does (it works but inevitably conflicts with RexxUtil), AND YOU CANNOT use a more than 8 character name for it, either, interpreter just can't find it, despite that "SysLoadFuncs" has more than 8 chars, that's internal and highly misleading. -- But I do now have a real, working, and even useful DLL that provides basic mouse function, all I need. With the supplied rexxmous.dll in \os2\dll you can test in Rexx: Run moustest.cmd from command line; should start making periods that show the loop is running. A careful long click with the mouse dead still should give position and button report, along with a large number for time stamp. Parse the returned string in Rexx as four words. For button 2 (right), you may want to turn off "Mouse Actions" from the VIO menu, but handily, it still passes the info through, just pops up the distracting menu. Button 3 works normally, but two at once isn't reliable, would require lengthy polling of status, so I've skipped it. You can build on the source code and add function. All you need is the Virtual Pascal compiler and rudimentary programming ability. Pascal won't let you go as far astray as C does, and operating the compiler itself is far less confusing; its "Make" is quite streamlined. Sadly, the Virtual Pascal site disappeared some time ago. But the last, Build 279 is still available for download from le généreux Pascalophiles at: http://pascal.developpez.com/compilateurs/vpascal/ Quick start for VP: The VIO and PM versions operate alike. I now use the PM version, shows more lines. Leave the compiler options alone unless QUITE expert; can easily get it to an inoperable state. But I advise turning off syntax highlighting and other unnecessarily colorful options of the editor. (Actually, I advise using any other plain ASCII editor you're familiar with whenever possible. The TP/VP editor has never pleased me and I avoid it until final stages of development.) Now, its file selector is also horrible (imitated from at least TP5), but you should be able to eventually find the files. Select "rexxmous.pas"; "Compile" menu, "Make", or just hit F9. If we both haven't messed up environment from the install, yours should report Success in the Message window. The new DLL will be in \VP21\out.os2, and must be moved to \os2\dll before test. That's about it. Turned out that I could do it, but had my doubts at times.