Jack - the trick is not so much reverse engineering the mapped files format as the J data-representation format. The good news is that I have already done this. If you have ACM access, you could take a look at my paper "J and Dyalog APL sharing resources: can't we all just get along?" in which I use memory-mapped files to pass data between J and Dyalog APL ( see http://portal.acm.org/citation.cfm?doid=602231.602247).
The trick was to write some APL code to parse the J array internal format, then read and write J arrays from and to the file. I was able to do this. You could figure out the J internal format yourself as I did, using the "3!:" family of foreign conjunctions, or wait until I can get home and post a copy of a paper I wrote detailing how I did this (this is briefly shown in the paper). I also have the APL code somewhere, if that would help - only some of it is in the paper. Let me know if you want me to dig up the APL code. Or maybe someone has already written C code to parse J arrays? Devon On 1/18/08, Jack Andrews <[EMAIL PROTECTED]> wrote: > > hi, > > is it worth my while reverse engineering > the jmf format? it feels like i should... > for example, if i've got a large number of > integers -- can't i just write them out and > have j map the file straight into a noun? > > for(i=0;i<N;i++) > fwrite(a+i;sizeof(long),1,f); > > what's the overhead in parsing such a file > (compared to J <-> J jmf)? > > ta, jack > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm > -- Devon McCormick, CFA ^me^ at acm. org is my preferred e-mail ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
