Its probably because the code sprawl is so vast, and so boring that no-one else wants to do it.Sometimes I wonder, is that really for any practical reasons, or is it to really stroke the vanity of the last few remaining guardians of "COBOL's legacy" ?
Indeed. And it is a really great tool for that. Rarely its brought up because analysis, documentation and testing is seen as "boring" and doesn't really massage this current "AI Hype" on i.e LinkedIn but really is the more realistic outcome. AI is already pretty good for generating tests to ensure the newbie humans are well guided.
Definitely for practical, bottom line reasons. Each mainframe hardware refresh cycle costs the companies that use cobol on mainframe many millions, yet that is still vastly cheaper than the huge cost of migration to the cloud, let alone the risk of business disruption. IBM just had a bumper year selling their latest versions of mainframe hardware and the associated consultancy to all the companies that they have locked in to using it, which is many of the really big firms, especially the longer established ones. It's still IBM's core business. They call it an "AI mainframe" now. Yeah, right.Sometimes I wonder, is that really for any practical reasons, or is it to really stroke the vanity of the last few remaining guardians of "COBOL's legacy" ?
www.tradingstocksnow.com
newsroom.ibm.com
Yes, AI has a great future in comedy I guess.I wonder how Elon's re-write of the US social security system's millions of lines of cobol into java is going? It seems to have gone very quiet...
Weirdly, they have had that strategy available to them for a decade in the form of outsourcing to india/china. They never really took it.
Or they are back at their old desk two month later for twice the bucks as a freelancer. That happend to a coworker of my dad.
it is not any of those things, though. see, for example, https://www.tue.nl/en/our-universit...-summaries-suitable-for-studying-and-research
Not that long ago. Has anything changed (ie, improved) at Boeing?
So technical documentation is quite a different thing to what you are referring to here.
our experience is that everyone says "they're good for (this one thing i personally don't want to do)" and then the research into that thing shows that they are not, in fact, good at that thing, and nobody can in fact agree on the one thing they're good for. this leads us to conclude that they are not good for anything.
Fortran has always been the preferred language of engineers, so it has acquired many useful features for engineering, including complex numbers and matrix processing. It also supports parallel programming using the MPI interface. It also has the KIND system, which makes it much easier to get the same results on different hardware. Rather than specifying REAL*4 or REAL*8, like we used to, you can tell it to give you a real variable with p decimal digits of precision, and and exponent range of at least r. The compiler will use that to select the appropriate kind of variable on your particular machine. A minor, but handy, feature is printing numbers in engineering notation, like scientific notation except the mantissa is greater or equal to 1 and less than 1000, and the exponent is a multiple of three (works nicely with kilo-, mega-, giga-, tera-, etc.).
It might be tricky finding a community for FORTRAN 77, because these day, most people are using Fortran 90, Fortran 2003, Fortran 2008, or Fortran 2018. FORTRAN 77 was the last Fortran to require fixed-format input, and there have been all sorts of improvements that make it more fun and useful to use.
True, like BLAS and LAPACK for linear algebra, and (my favorite) the Fortran Astrodynamics Tool Kit:
GOTO was deprecated in Fortran 90 and removed in Fortran 95, though some compilers probably still support it.
C and C++ have goto. Pascal and Object Pascal do as well. It's sometimes useful (Oh, no! Sacrilege!).
I still use the algorithms published in the Journal of Applied Statistics, mainly for the code quality and robustness - all the mathematical issues, convergence, handling of floating point, etc, have been carefully thought through.
later on down the page, they say:
