Stability vs Innovation
=======================

I have found that I do use pretty old software just because it works. It is
obvous that new versions can provide more features (but it is often
questionable if these innovations have actually a positive impact on actual
productivity..).

In the other hand, I do need to use the same program on several differen
computers (on a pocket computer, on a laptop and on desktop workstation) and
any differences in behaviour on in interface are very annoying. Another problem
is that newer software often does not compile on older systems. And the older
can mean even system with GCC few versions old or a slightly older Tcl/Tk. The
situation on a much older systems is much worse, of course.

Just a few example:

* the SC spreadsheet 6.21. It's the last version which is available for MS-DOS,
  various Unices (including IRIX), it is included in the OpenWRT for the Ben
  NanoNote, and it still can compile on the modern Linux; there is a 7.x line
  which has changed shortcuts, adds some complexity but no new usefull features

* the Bywater BASIC 2.20. Never ones do not compile on the Zauri, on IRIX
  systems and so 

* the Gtk+ 2.x library. The newer versions are too slow on the IRIX (if they
  compile at all) and I dislike the "newer" file dialog - I wish to use the one
  which was introduced in the Gtk+ 1.x line

* the Vim 6.x. I do not see any single feature in the 7.x and 8.x lines which
  approves use of a muc hslower and larger piece of software. Well, I use 7.0
  on some devices because they have enough power to use it.

Update: it's the bwbasic 2.20, not the 2.21. I'm probably blind...