Michael Grant wrote in a message to Bo Simonsen:

 MG>> There's already been some improvements in the Squish code, and I 
 MG>> am hoping that there will be a removal of the Squish read/write 
 MG>> buffer limit for 32-bit environments. I sure could use an 
 MG>> improved version of Squish/Win32 right now, as my system is pretty
 MG>> complicated and I have quite a ways to go before I can switch 
 MG>> everything over to Linux.

 BS> By boss is a old squish user, he told about the buffer problem,
 BS> if a line
 BS> had more than 512 charecters, Squish didn't read the rest.. And it was
 BS> very problematic because he did have over 200 points, and many links
 BS> too..

 MG> The upper write limit is 128k in the 32 bit version. Isn't that a
 MG> bit more than 512 characters?

 MG> Hpt has proven that Squish doesn't need this limit in 32-bit
 MG> environments, because it's Squish compatible and it doesn't have
 MG> the limit. It'll process anything as long as you have enough memory
 MG> on your system.

Heh.

This reminds me of a few years back,  I was involved in this othernet,  and
somebody in there was running software that would,  from time to time, 
crash rather spectacularly.  Not only would it crash,  but when it did, 
the system would somehow end up with a packet file in the outbound and that
file was *huge* to start with.  I remember getting hit by one that was over
5 meg!  If the sysop didn't catch the problem,  it'd somehow recirculate
and the next copy would be bigger,  and bigger,  and...

Biggest one I ever saw here was 8M+.  It's a wonder and a testament to how
robust most of the software in use is that the mail handling and so forth
along the way didn't choke on this stuff.

---