From: Gregg Reynolds (unicode@arabink.com)
Date: Thu Jun 30 2005 - 10:41:07 CDT
Erkki Kolehmainen wrote:
> Although I wasn't supposed to see the exchange, I'd like to comment on
> the following statement:
> "After all, ask yourself why legacy compatibility was required in the
> first place. Maybe many reasons, but one of them was surely so that
> software that uses a legacy encoding internally can continue to function
> without modification, with only an import/export filter."
>
> In my mind, legacy compability was required to ensure that the data that
> has originally been encoded using whatever scheme remains processable,
> often but not necessarily together with data originally encoded using
> whatever other scheme. A reliable way to convert the data (into Unicode)
> in order to preserve it permanently was required, not the preservation
> of some pieces of software except possibly for a transition period. In
> fact, many of the platforms for the implementations that were originally
> used to process and store the legacy data were becoming extinct already
> in the early days of Unicode and many more are extinct by now.
It would be interesting to see real data. I'd be quite surprised if
there weren't billions of $ worth of legacy encoded data sitting around.
I wonder how many Japanese financial institutions have converted their
legacy data to Unicode? As for processing, I recall reading somewhere
around the time of the y2k rush that worldwide investment in COBOL is
over a US$ 1 trillion. Even if it's not that high, it is surely multple
billions. I can tell you from personal experience that lots of large
businesses run COBOL programs every night for which the source code has
long since been lost. I'm also sure there are lots of mainframe
programs written in S/370 Assembler - I once worked briefly at a bank
processing center where the entire system was written that way! Those
systems will probably never convert to Unicode. If you have legacy
data, and legacy programs that work, what incentive is there to convert?
-g
This archive was generated by hypermail 2.1.5 : Thu Jun 30 2005 - 10:42:11 CDT