> Scientific code is often write-once, run-once, done. If only it were so! Lots and lots of "scientific" code turned out to be re-formatting data. There were

> Scientific code is often write-once, run-once, done. If only it were so! Lots and lots of "scientific" code turned out to be re-formatting data. There were some things I wrote in Perl, then re-wrote in Java (which was the standard for non-compute-bound code at the time). But then there was the actual computation... Holy cow the code from the 1970s we had to learn, to transform into freaking Java code. And holy cow the poor souls that came after me, howling in torment at my code.