[Info-vax] RMS internals?

[Arne Vajhøj]

Howard S Shubs wrote:
> Not specifically.  I'm reading records of whatever length using block 
> reads and RAB64 to get larger buffers, so I need to parse out the 
> records myself.  The record length I'm specifically working on is over 
> 33000 bytes long.

In that case it is not a really a valid file.

See:
   http://h71000.www7.hp.com/doc/731final/4523/4523pro_006.html#fab_mrs

> But I figure it'd be a Good Idea to implement every record type I can.  
> Given the link sent earlier, I'm a little concerned about other possible 
> implementations of VARIABLE records.  I'll check that out on Monday to 
> make sure I've covered remaining bases.

The variable format is welldefined:
   2 bytes with length
   data butes
   0 or 1 bytes to fill to even number of bytes

> I find it kind of funny.  The records I'm receiving alternate.  I get 
> one record of 32767 bytes, then a record of 330 bytes, completing the 
> *actual* record length.  It repeats until I have all the records.  Fun!  
> I'm not assuming it'll always work that way, note.

If you have an invalid file, then it is not surprising to see
weird results.

Arne