STL Time Machine Report #24 - Sunday 12 July 1998 (1998-07-12) The "stupid windowing trick" I mentioned in TMR #23 seems to have occurred in several places. They found it one more time in a re-test. This is the problem where the transaction date is windowed using the year from the current date, which causes an entire file to have dates in the year 2099. They will still have time in the final cycle of group 3 testing to verify the corrections. I had an interesting conversation with a Unix propeller-head last week (it's okay, I'm a dinosaur mainframe pocket-protector geek). He laughed when he told me this story, because the development group insisted their application "didn't do anything with dates". They found an unexpected problem in the time machine with a daily file. The timestamp is converted to a character string YYYYMMDDHHMMSS. This is compared to the timestamp of the previous day's files, to prevent an older file from overlaying a more recent update. Now I only have a slight understanding of C, but it seems the two timestamps were being compared as 32-bit int's, and since the numbers exceeded 12 digits, the conversion was failing, resulting in negative numbers that just happened to compare correctly, for 1998 dates, but did not work in the time machine with the rollover to 2000. The fix for this is to compare the two timestamps as character strings. This problem was completely unexpected. I think it's a big win, because this problem was found and will be fixed well before it could cause any damage. More on the "Summer of Upgrades" We're starting to install Unix hardware & software upgrades. Nearly half of our CICS regions have been upgraded to Release 4.1. In August, they're upgrading to faster engines and more modern disk drives on our mainframe. Our technical services department is also working on upgrading from MVS/ESA 5.2.2 to OS/390 by September. It seems we've done a mainframe OS upgrade every year for the last four years. Last Monday, they completely re-installed NT 4.0 on my PC at work. Some kind of problem with the 16-bit support prevented me from entering my timesheet. I lost almost a whole day for that. COBOL Had an interesting conversation with a programmer who works with a really old application. This is the one that had the most problems with missing source when we first assessed our code back in late 1996. All the source was found and the biggest problem was Panvalet ++INCLUDE's that were stored on test libraries rather than production libraries. This old system only has a few users, but they like it and don't want to get rid of it. It's all being converted to COBOL II. I think it's good that we're finally getting off OS/VS COBOL. Most of the problems converting to COBOL II have been related to statically linked subprograms and NORES/NCAL compile/link options. It will be much easier to upgrade fromo COBOL II to COBOL for MVS (OS/390) and VM. High-Level Assembler. Seven old assembler programs were converted to HLASM, and will be tested in the Time Machine. Miscellaneous Notes. It looks like our third party database will be installed with the Y2K and LE/MVS compliant version this month. It still looks like it will be first or second quarter 1999 before we can install Language Environment/MVS as the default runtime libraries. Our QA department is staffing up. Test script development is underway. Oil Refineries There was some interesting discussion in c.s.y2k about this in the last week. I know one refinery that is essentially self-sufficient in electricity, because the burn methane for their own generators. I have no idea if their generating system is Y2K compliant. The Wood River refinery near St. Louis has no backup electrical generation capacity. Electric Utilities Mickey reported some good news this past week. I will feel much more confident if I start hearing stuff like Mickey's reports from our local electric utility, AmerenUE. Interesting URL's. Here they are again, or you can find most of them at: http://home.att.net/~arnold.trembley/y2klink.htm IBM tips on building an OS/390 Time Machine: http://www.software.ibm.com/year2000/y2ktestsystem.html IBM's free download, the Time Machine for CICS: http://www.software.ibm.com/ts/cics/txppacs/ch15.html Microsoft patches for Windows 3.1: http://www.microsoft.com/ithome/topics/year2k/product/wind31.htm Microsoft patches for Win95: http://www.microsoft.com/ithome/topics/year2k/product/win95.htm Windows NT 4.0 Workstation fixes: http://backoffice.microsoft.com/downtrial/moreinfo/y2kfixes.asp NT 4.0 Server, standard & enterprise editions: http://www.microsoft.com/ithome/topics/year2k/product/WinNt40SvrUS.htm AT&T's Year 2000 page: http://www.att.com/year2000/ The entire collection of Cory Hamasaki's DC Y2K Weather Reports: http://www.sonnet.co.uk/muse/dcwrp.html As of 1998-07-12, My countdown now reads: 112 days until 1998-11-01 (Beta Test begins) 173 days until 1999-01-01 (External Testing begins) 538 days until 2000-01-01 (Rollover) There's still a lot of work to do, and there's still time to fix code and make contingency plans. Previous Year 2000 Time Machine Reports are available at: http://home.att.net/~arnold.trembley/tmr.htm STANDARD DISCLAIMER: I am NOT an official corporate spokesperson. My opinions should not be held against my benevolent employer. -- Arnold Trembley http://home.att.net/~arnold.trembley/ "Y2K? Because Centuries Happen!"