Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Path: utzoo!mnetor!uunet!husc6!bbn!rochester!cornell!batcomputer!braner From: braner@batcomputer.tn.cornell.edu (braner) Newsgroups: comp.sys.atari.st Subject: Megamax 2.0 aka Laser C Message-ID: <3085@batcomputer.tn.cornell.edu> Date: Fri, 4-Dec-87 12:50:07 EST Article-I.D.: batcompu.3085 Posted: Fri Dec 4 12:50:07 1987 Date-Received: Wed, 9-Dec-87 04:40:12 EST Reply-To: braner@tcgould.tn.cornell.edu (braner) Organization: Cornell Theory Center, Cornell University, Ithaca NY Lines: 58 Summary: upgrade time! [] Just got issue no. 2 of "printf", the Megamax Newsletter, announcing their "lastest" (sic) product. Here are the news in brief, and then a few inquiries. Megamax C version 2.0 (now called Laser C) will be available soon. If you have Megamax 1.1 you can send your original disks + $20 (a check made to Megamax), and they will send you the new version (including the brand new manual) as soon as it is ready. The changes in new version include: Better graphical shell, including a "STDIO" window for command-line die-hards (like me :-) and a "FILER" tool that is friendlier than the desktop for file copy/rename/delete etc. A RAM cache that automatically uses the part of RAM that is not used by the running programs. A MAKE facility. No more 32K limits on code segments or extern vars. (Absolute referencing - bigger, slower code?) (And what about local vars larger than 32K?) DRI compatible linker, and faster than the old one. Some sort of debugging monitor. Faster and more accurate (read: less buggy) floating point. My comments: Can one now adjust the stack size an application gets without recompiling the startup.c (and rearchiving syslib)? (I redid the old startup code so that I can now do: "long _stacklen = 123456; main(){...}" ) I have had the opportunity to try out the new FP package. It is very fast (faster than Absoft FORTRAN). But did they improve the calling procedures? I.e., is the C statement "a=b+c; (all doubles) translated to a simple "jsr _fpadd", or is it still routed through a big jump-table lookup scheme? With the faster FP package, this can introduce almost a factor-of-2 slow-down! If, when debugging a program, the thing locks up and a RESET is necessary: does the RAMcache (and the latest version of the source code, if compiled-and-run from the editor, as is now possible) get lost? Since I don't have a hard disk, maybe I'll stick with the trusty RESET-proof RAMdisk. Disclaimer: I am not affiliated with Megamax. - Moshe Braner