"Gabor Szabo" schreef: > {Cobol etc.] > IMHO - and I really saw only a few such companies - these companies > have 0 automatic tests so it would cost them a lot of time and money > to test their application on the new and shiny Cobol compiler.
I once worked on tests for a national center of a bank, to monitor the nightly runs of big money flowing. The Cobol programs ran on mainframes; there were several IBM PCs, with character terminal emulation software, connected to them. On each PC there was a stack of C-programs, generated with yacc and lex. When a fresh screen came up, a main program deducted the proper test for it and ran it. When the test failed, the PC would start beeping and flashing, to wake up the one operator that remained: this software replaced a couple of people that once did the same tests with mental arithmetic and a pocket calculator. The terminal emulation software had a very raw INT-something-API (HLLAPI IIRC), and IBM had updated the terminal emulation software but hadn't updated the (never supported) linkable library layer for the API, so the first thing I had to do was to create a replacement library. That took only half a day, because the documentation of IBM was OK, and I could generate much of the C-code with a few macros. I wondered why IBM hadn't done that, it would have been a nice gesture. http://www.webopedia.com/TERM/H/HLLAPI.html Next I had to check the test programs, because they failed often. These were full of strdups without frees and things like that, they were made by a Pascal programmer. It took me two days to fix all those yacc/lex sources, throwing away about 90% of them because there was a lot of duplication. Basically such a program would check the layout of the screen, copy a few numbers from it and do some calculations on them, then store the result in a log. And of course it would bark when things didn't add up. Monitoring, regular expressions, character based: nothing much has changed. http://search.cpan.org/~grommel/Convert-IBM390/IBM390.pod http://search.cpan.org/~ingy/Inline/Inline-API.pod "The Inline community will decide if your implementation of COBOL will be distributed as the official Inline::COBOL or should use an alternate namespace." :) -- Affijn, Ruud "Gewoon is een tijger."