Front page | perl.perl5.porters |
Postings from November 2007
On Nov 22, 2007 3:13 PM, Michael G Schwern <schwern@pobox.com> wrote: > But still, one might argue, that leaves a whole lot more Windows developers > than the 1% which said they've touched VMS in the last year. Due to no one's fault but my own, the survey was not announced in places where people who use Perl and VMS were likely to see it unless they were also pretty deeply plugged into general Perl news sites. It was not even posted to vmsperl AT perl DOT org, where Ask tells me there are 139 subscribers (which is about 100 more than I expected). There are assorted VMS forums where questions about using Perl or solutions written in Perl come up with some regularity, and had I been paying better attention, I'm sure we could have reached quite a few more people. Next time, give me a nudge to help get the word out. The percentage would still not be large, but it would be interesting to know more about what people are using Perl for in the wild. > Yes, but consider the following... > > It takes a whole lot more developer effort to keep Perl running on Windows > than on VMS. VMS has a relatively unified installation. One compiler, one > build system. Once you get something working on one VMS machine, you're > pretty much done. It doesn't change very often or very much. > > Windows, in contrast, has a very varied and fractured installation base and > tool chain. For operating systems, there's 98 and XP and 2000 and Vista. And > then all the various subtly incompatible versions of Visual C++ or maybe > they're using MingW or Borland or gcc? Old nmake? New nmake? dmake? And > let us not get into Cygwin. Well, there are certainly wrinkles involved in supporting three hardware architectures, two filesystems, and assorted versions of UNIX compatibility (or lack thereof) in the C run-time. But compared to Microsoft platforms, yeah, once things are working on VMS they tend to stay that way. The currently shipping C compiler for VAX runs on VMS versions going back to v5.5-2, introduced in 1991. The very first VMS system, the VAX 11/780 introduced in 1977, can run VMS versions as late as v6.2, introduced in 1995. So there's no reason in principle that we couldn't build a very recent version of Perl, possibly even 5.10, on a 30-year-old board system (you heard me right, the 11/780 did not have a microprocessor). We'd have to find somebody still running one that had adequate memory and disk space and was willing to donate (I'd guess) a month or two of processing time. Or you could prove it's possible using the 11/780 emulator in SIMH (http://simh.trailing-edge.com). It would be completely useless as well as completely cool, but it would drive home your point that getting something working on VMS usually means it's still working a decade or three later, and Jarkko's point in another thread that Perl's extreme portability means it's well-poised to run on anything that comes along. I doubt the 11/780 experiment will happen (too many other things to do), but if it does, I have dibs on the article title: "How to Build Perl Without a Microprocessor."Thread Previous | Thread Next