mwm@violet.berkeley.edu.UUCP (01/26/87)
Ok, time to make rude, nasty noises about Lattice 3.10. Those who hated my time doing the same for MANX can consider this restitution. Those who haven't upgraded to 3.10 can take this as a warning. First, let me point on that only for large systems do I have problems with Lattice. The smaller things I've been playing with (one to be posted in a few days....) have compiled and worked like a charm, with either the large or small model code. But Lattice did break the semantics of <> includes. However, the two large programs (the release of MicroGNU with the browser "can't-back-up" bug fixed, and vt100 2.4) I'm playing with compile (after a little tweaking), and then die in mysterious ways. So, _the_ question: Has anyone gotten the release version of mg1a (on Fish disk 42) or vt100 v2.[24] to compile & run under 3.10? If so, what have you done to make it work? Symptons: on mg, line deletion/insertion change the "top-line" in the wrong way, and block deletion runs the system out of memory then hangs everything. The EXACT same code works like a charm with 3.3b. With vt100 2.4, it crashes with an "Address Error" Guru on the way up. Examining it with MetaScope shows that it goes into the second call to fopen in InitDefaults, and then never comes back (the first call is not made, as there are no arguments). My suspicion is that Lattice 3.10 has problems in the I/O library, *if* the resulting program is large. Can anyone confirm this? Has anyone compiled something "large" and had it work? Thanx, <mike