Subscribe by Email
I apologize in advance: this post is a bit of a rant.
I’m tired of hearing, at work (and not just here, but in my previous jobs as well) the bullshit about “supported versions” – that some piece of software only works (or is “supported”) with some particular older, non-upgradable version of a library, a compiler or interpreter, a Linux distribution, and so on.
To me, that just means that the software is badly programmed, that it was compiled by trial and error, using hacks, compiler/library bugs, and non-documented tricks, by people who don’t really know what they’re doing, until it… “seems to work… mostly”. Then, they took note of the versions of everything around the software – OS version, libraries and so on – and, then, say that only that exact configuration was “supported”.
Sometimes, it gets so bad that you can’t even install security updates – which fix bugs, but don’t upgrade stuff to newer versions – because it’ll become “unsupported”, and, quite often, begin to “crash and burn” often.
The shocking part about this is that everyone around me seems to believe that all this crap is normal, instead of simply refusing to use software which requires Perl 5.8.3 and breaks with 5.8.4, or Java 1.4.2 but not 1.5.x, or that specific, unpatched, insecure version of glibc.
It’s almost as if those people – who, amazingly, work in IT! – still believe that software is some ungraspable, unknowable, “magical” thing, that works (or doesn’t) as if by magic.
People, unless the newer version of a library or interpreter actually has a bug which makes some older app not work, if the app stops working, it’s the app’s fault. Stop putting up with this idiocy.
And yes, in 99% of these cases, it’s proprietary, ultra-expensive software. For some reason, this almost never happens in Open Source…
No related posts.
LOL! Oracle software yes?
It’s one of the best examples, yes.
Though what inspired this post today was when a co-worker talked about some (bought) application which only worked with a very old Perl version (I don’t think the software itself was in Perl, but it used it for something…).
There is a world of difference between “working” and being “supported”.
Nuno: I know that. In this last case, it was really a question of working. And I’ve seen dozens of cases where it seemed to work, but would crash, and crash, and crash, and crash, without any “normal” reason – just because it wasn’t using the exact library version (including minor version) they were able to compile it with.
“Support”, as in, “you pay us thousands of dollars a month, and, when it breaks, you can call us”, is another matter – but, IMO, as inexcusable as the first.
It’s acceptable to say “requires version X.Y.X or later”. It’s not acceptable to say “requires version X.Y.Z, period”.
Indeed the software is mostly written in Perl and C using Motif graphics interface and, YES, ORACLE database. But the point here is that what crashed was a custom (customer specific) script written in Perl, long before the actual Perl interpreter version exists. So that might be the case that newer Perl interpreter doesn’t run the script that used to run fine with the previous Perl interpreter version… maybe the problem then is not on the miswritten script, because it used to work with previous Perl interpreter and now it doesn’t anymore…