mpd@anomaly.sbs.com (Michael P. Deignan) (06/14/91)
Many of you will remember reading two weeks ago about the problems I was having with Clipper 5.01 applications (ie: mysterious program "hangs", random errors, nonsense errors like: "unrecoverable error 415: unable to open overlay file '<happy face>', etc.) which, although intermittent, were regular enough that I eventually had to scrap 5.01 entirely. At that time, I stated that the problem seemed to be .RTLink related. Other comp.database users, spouting the Nantucket line, claimed "oh no, it *HAS* to be your program." Well, late last week, enter the latest version of Blinker, v1.5. At this point, I decided to dust off the Clipper 5.01 release and give it another try. The results were astonishing. I re-compiled all source code with the v5.01 compiler, and using Blinker and .RTLink, I created two binaries, "CROSBY1.EXE" (blinker) and "CROSBY2.EXE" (.RTlink), and executed both on the same machine for two days. Note that these were binaries created from the SAME OBJ files - simply by calling a Blinker link script one time, and then the .RTlink script immediate afterwards. The results? You guessed it! CROSBY1.EXE did not experience any "odd" problems. The CROSBY2.EXE binary, using the same keystrokes just entered in the CROSBY1 program, would suddenly "hang", terminate with our favorite happy-face error message, begin to re-execute the program from the root module again (interesting since the program only executes the root module once, and only once, and never returns to it!) among several other interesting occurences which ultimately rendered the CROSBY2.EXE binary useless for all practical purposes. I now have over 20 copies of a v5.01/Blinker executable being tested in the field (since monday) and have had *NO* reports of "odd" lockups, etc. So, was it "my code"? Well, I guess so, if Clipper programmers will be able to state in the future with complete support from Nantucket that the Clipper code some programmers develop is so complex that even the "supplied" linker can't link it correctly.... Replies or followups to anomaly!mpd@quahog.ssd.ray.com. MD -- -- Michael P. Deignan / -- Domain: mpd@anomaly.sbs.com / This space intentionally -- UUCP: ...!uunet!rayssd!anomaly!mpd / left blank... -- Telebit: +1 401 455 0347 /