rpk@rice-chex.ai.mit.edu (Robert Krajewski) (10/03/90)
I'm in the middle of switching over from another (*ahem*) DOS C++ compiler to Zortech 2.1, and so far I'm pretty happy. However, I've noticed the following problems as I've been bringing over Windows code: 1. _huge is not supported. This doesn't affect us -- yet, anyway. 2. windows.h defines a macro LPTR (used with LocalAlloc), and Zortech defines this when in large or compact models, meaning something entirely different. I'm surprised that nobody has pointed this out in either comp.lang.c++ or comp.windows.ms. This isn't really a horrible bug, but doesn't the ANSI spec say that implementation-dependent identifiers should start with _ ? SPTR and LPTR ought to be renamed _SPTR and _LPTR, or maybe even something else more descriptive... 3. When errors and warnings are encountered in included files, it would be nice if Zortech printed out the full path of the included file. For some reason, my copy of Epsilon is tricked by the relative paths it's getting (during the C-X M (make) command). 4. -o, to specify where the object file ought to go, doesn't seem to obey the given extension. For example, with -ofoo.obl in the command line, the object file still goes to foo.obj. I notice that ztc2br gets -ofoo.obj when this happens. (I'm using the protected mode version.) 5. If any phase of the compiler fails, ztc.com still tries to run the subsequent ones. It's hard to beleive that a product as sophisticated as this exhibits such autistic behavior. 6. And although the verbosity of running the phases helped me detail the problems in (4), most of the time, I would like to be fooled into thinking that ZTC is just one program and not see the internal commands that are being executed. 7. Zortech's MAKE doesn't seem to want to execute .BAT files ! ---- Robert P. Krajewski Internet: rpk@ai.mit.edu ; Lotus: robert_krajewski.lotus@crd.dnet.lotus.com