Path: utzoo!utgpu!news-server.csri.toronto.edu!helios.physics.utoronto.ca!ists!yunexus!shields From: shields@yunexus.YorkU.CA (Paul Shields) Newsgroups: comp.sys.amiga.programmer Subject: Re: Manx vs. SAS/Lattice Message-ID: <22394@yunexus.YorkU.CA> Date: 13 Apr 91 16:57:22 GMT References: <9104021420.AA10848@thunder.LakeheadU.Ca> <1991Apr4.034920.16298@marlin.jcu.edu.au> <1991Apr4.180217.19773@nntp-server.caltech.edu> <1991Apr5.030228.28756@marlin.jcu.edu.au> <966@boing.UUCP> Organization: York U. Computing Services Lines: 56 kuch@reed.edu writes: >I'm a Manx owner who recently received the 5.0d upgrade. In light of the major >and crippling bugs in 5.0a and 5.0b, I have some apprehension about getting >started with it. Does anyone out there have any experiences to report? Sure, I've been using 5.0d for the past 4 or 5 months on a full-time project, so I've had lots of time with it. It works quite well for me. Here are the problems I have encountered: 1) There are a couple annoying problems with sdb, all of which have fairly easy workarounds... - If you quit before single-stepping at least one instruction, sdb may hang. - If you resize sdb's window before it is finished loading your program, you sometimes crash. - If you distribute source modules throughout several directories, realize that sdb will only look in the current directory for source modules. I found this really annoying because I put source in various places to help manage complexity. 2) If you use any floating point math functions, you'd better make sure you put the -lm option before the -lc option on the link command. The documentation says that the non-math printf library will print "f" or "g" when you try to print floats, but my actually it silently fails. 3) When linking with more than 64K of initialized data, the linker complains and there seems to be no way to get it to work without placing the data in the code segment. I am uncertain whether that is satisfactory, but it works. 4) The optimizer is really inadequate. This is why I'm about to take the plunge and try using gcc to optimize instead. >I'm also wondering if Manx has been left behind in the functionality race. >Can any users out there give me a run through of the highs and lows of each >package as they see them? I haven't used Lattice, but Manx appears to be fully ANSI compliant. (If any of you know better, do tell.) What functionality would be nice? I have used the Microsoft codeview debugger on PC's, and think it would be a "good thing" (tm) for the Manx debugger to watch variables, ie: to display the values of a list of variables at each single-step automatically. An execution profiler would be nice. It's time for me to start tweaking the speed of my code up.. a profiler it would be _so_ helpful since the optimizer doesn't work well enough. -- Paul Shields