Thread View: comp.sys.mac.programmer.codewarrior
11 messages
11 total messages
Started by Steve Roy
Thu, 19 Oct 2000 00:00
Mac Java debugging
Author: Steve Roy
Date: Thu, 19 Oct 2000 00:00
Date: Thu, 19 Oct 2000 00:00
40 lines
2437 bytes
2437 bytes
Hi, I am, to say it nicely, pretty upset with the state of Java debugging in CodeWarrior 6 on Mac OS 9. Debugging is waaaaaayyyy to slow, and the unsupported Metronub debugging displays garbage instead of correct names in the Variables pane. This means that the official debugging is unusable. I have tried it on many machines. My main machine is a 9600/300. There I give Metrowerks the benefit of the doubt, it's an older machine. But, I also tried it on a G3/233 and a G4/400, and it's still unusable. I know Metrowerks know it's slow and it will be better on Mac OS X, but what we need is a real solution right now. And then the unsupported Metronub debugging. Why are variable names mangled? I know it's unsupported, but goodness, you guys probably tried it. If you saw that variable names were mangled like that, why the heck did you put it on the CD? It all comes down to: what are our options? We are writing software for Mac OS 8/9. Although we have installed Mac OS X, this is still a beta OS and we are not going to move our development machines and build machines to OS X until it's final. This is just common sense. At the very least, we will be writing software on Mac OS 8/9 for another 6 months. What are we supposed to do in the meantime to debug our code? And don't tell me about putting System.err.println() statements in our code. This is useful, but is not a debugging facility. Metrowerks is a serious company and as such should have devised a serious plan to help their Mac users. Anyone has seen this mangled variable names problem? Am I missing something, yet again? Steve -- Steve Roy <sroy@mac.com>
Re: Mac Java debugging
Author: Eric Albert
Date: Thu, 19 Oct 2000 00:00
Date: Thu, 19 Oct 2000 00:00
29 lines
1838 bytes
1838 bytes
In article <B61484D4.A1B3%sroy@mac.com>, Steve Roy <sroy@mac.com> wrote: > I am, to say it nicely, pretty upset with the state of Java debugging in > CodeWarrior 6 on Mac OS 9. > > Debugging is waaaaaayyyy to slow, and the unsupported Metronub debugging > displays garbage instead of correct names in the Variables pane. The speed isn't really Metrowerks' fault, or that of the Java team at Apple. Sun's debugging API reflects Sun's belief that TCP/IP is the best solution for everything, which isn't really always the case. In this case, you've got two processes running on an OS with pretty poor process management and a networking architecture that doesn't appear to be optimized for communication between local apps. Those are two things that are much improved on Mac OS X, fortunately. Though I don't have much experience with the CodeWarrior debugger -- println's and the mrj dcmd can actually accomplish quite a lot if you need them to :) -- I suspect that remote debugging may perform reasonably close to local debugging, if not better, since you don't have the overhead of the process switching. Give that a shot if you can set two machines up next to each other. -Eric -- Eric Albert ejalbert@stanford.edu http://www.stanford.edu/~ejalbert/
Re: Mac Java debugging
Author: Steve Roy
Date: Thu, 19 Oct 2000 00:00
Date: Thu, 19 Oct 2000 00:00
18 lines
1449 bytes
1449 bytes
Eric Albert <ejalbert@stanford.edu> wrote: > The speed isn't really Metrowerks' fault, or that of the Java team at > Apple. I am not blaming either of those for this problem. But I am blaming Metrowerks for not acknowledging that there is a problem and for not providing a workaround. At the very least, they should have assigned one engineer to the Metronub plugin to ensure minimal functionality, while acknowledging the limitations of the workaround. For now, there is no workaround, except for... println() statements. What year are we in? Steve -- Steve Roy <sroy@mac.com>
Re: Mac Java debugging
Author: Robert Gulian
Date: Thu, 19 Oct 2000 00:00
Date: Thu, 19 Oct 2000 00:00
29 lines
2187 bytes
2187 bytes
In article <ejalbert-3FBDD1.13140019102000@nntp.stanford.edu>, Eric Albert <ejalbert@stanford.edu> wrote: > The speed isn't really Metrowerks' fault, or that of the Java team at > Apple. Sun's debugging API reflects Sun's belief that TCP/IP is the > best solution for everything, which isn't really always the case. Well, debugging Java on Windows is fast and Microsoft isn't tooting Java as the official programming language of it's new operating system like Apple is. Context switching doesn't seem like an adequate reason for the slowness. We're talking MINUTES to display all the member variables of "this". A context switch on a PowerPC is faster than a context switch on an Intel machine. I put the blame squarely on Metrowerks and Apple's shoulders. If Apple wants folks to use Java on their next OS, they better make sure they can use it the current one. If Metrowerks wants to charge hundreds of dollars for a Java development environment, they better make sure it's usable. Sun is just a contractor in this development effort. I'm a java newbie but an old Mac development hand. Yeah, I can use macsbug and println but I'm sure not going to be able to convince my co-workers that the Mac is a cool machine to develop Java on. And that should be Apple and Metrowerk's aim. Bob bgulian@staff.iwon.com
Re: Mac Java debugging
Author: Steve Roy
Date: Thu, 19 Oct 2000 00:00
Date: Thu, 19 Oct 2000 00:00
21 lines
1506 bytes
1506 bytes
Steve Roy <sroy@mac.com> wrote: > And then the unsupported Metronub debugging. Why are variable names mangled? > I know it's unsupported, but goodness, you guys probably tried it. If you > saw that variable names were mangled like that, why the heck did you put it > on the CD? Let me emphasize this. Does anybody else see the mangled variable names in the debug window when using the Metronub debugger? I don't remember seeing reports of anything like this in this newsgroup. But I see this on two machines here and my boss sees the same thing at the head office. I *have* to be missing something. I sent a screen shot to Metrowerks tech support. Steve -- Steve Roy <sroy@mac.com>
Re: Mac Java debugging
Author: Burt Johnson
Date: Fri, 20 Oct 2000 00:00
Date: Fri, 20 Oct 2000 00:00
20 lines
1445 bytes
1445 bytes
Steve Roy <sroy@mac.com> wrote: > Steve Roy <sroy@mac.com> wrote: > > > And then the unsupported Metronub debugging. Why are variable names mangled? > > I know it's unsupported, but goodness, you guys probably tried it. If you > > saw that variable names were mangled like that, why the heck did you put it > > on the CD? > > Let me emphasize this. Does anybody else see the mangled variable names in > the debug window when using the Metronub debugger? > No, I've never seen that problem. Of course I use the unsupported metronub debugger, because the Sun debugger is too slow to be usable... -- - Burt Johnson MindStorm Productions, Inc. http://www.mindstorm-inc.com
Re: Mac Java debugging
Author: MWRon
Date: Fri, 20 Oct 2000 00:00
Date: Fri, 20 Oct 2000 00:00
42 lines
2151 bytes
2151 bytes
In article <B614D7EF.A1E0%sroy@mac.com>, Steve Roy <sroy@mac.com> wrote: >Eric Albert <ejalbert@stanford.edu> wrote: > >> The speed isn't really Metrowerks' fault, or that of the Java team at >> Apple. Are you using Carbon Lib 1.1 this really slows it down. >I am not blaming either of those for this problem. But I am blaming >Metrowerks for not acknowledging that there is a problem and for not >providing a workaround. What do you think MetroNub is if not a work around. We have also added features like the variables to "none" option and other things to make this faster. I don't think it is right to say we are ignoring it or aren't doing anything at all. We are trying hard to fix this. > At the very least, they should have assigned one >engineer to the Metronub plugin to ensure minimal functionality, while >acknowledging the limitations of the workaround. Read the release notes on MetroNub. >For now, there is no workaround, except for... println() statements. What >year are we in? I've had one user send me a photo of the mangled names in Java using the MetroNub the others don't seem to have this problem I'd like more information. This may be a problem an extension conflict. I don't mean to be unsympathetic I just want to help and I need information to do that. Ron -- "...we feel it's truly the best of the bunch." Internet Week Reviews...Java Wars 2000 http://www.internetwk.com/reviews00/rev100200.htm Metrowerks, "Software Starts Here" - Ron Liechty - MWRon@metrowerks.com
Re: Mac Java debugging
Author: Steve Roy
Date: Fri, 20 Oct 2000 00:00
Date: Fri, 20 Oct 2000 00:00
30 lines
2036 bytes
2036 bytes
MWRon <mwron@metrowerks.com> wrote: > Are you using Carbon Lib 1.1 this really slows it down. I just checked, I have CarbonLib 1.0.4, so this is good. I also tested with the Sun debugger and the automatic variables option. It speeds up stepping (although it's still not as fast as Metronub is) but it doesn't help at all with expanding objects to look at its variables. > I've had one user send me a photo of the mangled names in Java using the > MetroNub the others don't seem to have this problem I'd like more > information. This may be a problem an extension conflict. > > I don't mean to be unsympathetic I just want to help and I need > information to do that. OK. Thank you. I myself sent a screen shot to Metrowerks tech support thursday night. I am willing to give you all the information in the world if this helps in locating the cause of this problem. At least now I know Metronub works for some people, so I know I can get it to work for me too. I'm sending you a full Apple System Profiler report via direct email. Steve -- Steve Roy <sroy@mac.com>
Re: Mac Java debugging
Author: Burt Johnson
Date: Sat, 21 Oct 2000 00:00
Date: Sat, 21 Oct 2000 00:00
23 lines
1954 bytes
1954 bytes
Steve Roy <sroy@mac.com> wrote: > I am willing to give you all the information in the world if this helps in > locating the cause of this problem. At least now I know Metronub works for > some people, so I know I can get it to work for me too. Just a note from a very impressed user. I haven't seen the problems you describe, but have seen a few myself. I have found MW tech support to be absolutely the best in the industry, with people jumping through hoops to fix problems for me when it was critical. All it seems to really take is being able to describe a problem completely and accurately, and to provide a project that exhibits the problem. I've used tech support for a lot of companies, and some made me literally scream and throw their junk in the trash. MW is a blessed exception -- their stuff usually works, and when it doesn't, they respond better than anyone has a reason to expect. Stick with them, and give them what they ask for. You'll end up making it better for all of us -- and solving your own problem in the process.:-)
Re: Mac Java debugging
Author: Anon Anonymous
Date: Sun, 22 Oct 2000 00:00
Date: Sun, 22 Oct 2000 00:00
43 lines
2714 bytes
2714 bytes
I only used C++ under CW 5.3. Now that I am working on Java 1.1 stationary projects on the Mac with CW Pro 6.0, I can see the problem people are experiencing. If the speed issue is new to CW 6, and did not exist in 5.3, then I would think the problem lies with CW 6, and not Apple, Mac OS limitaitons, etc. Are ther any 5.3 users having problems with slow java debuggers? Cheers, -Navid. In article <ejalbert-3FBDD1.13140019102000@nntp.stanford.edu>, Eric Albert <ejalbert@stanford.edu> wrote: > In article <B61484D4.A1B3%sroy@mac.com>, Steve Roy <sroy@mac.com> > wrote: > > > I am, to say it nicely, pretty upset with the state of Java debugging in > > CodeWarrior 6 on Mac OS 9. > > > > Debugging is waaaaaayyyy to slow, and the unsupported Metronub debugging > > displays garbage instead of correct names in the Variables pane. > > The speed isn't really Metrowerks' fault, or that of the Java team at > Apple. Sun's debugging API reflects Sun's belief that TCP/IP is the > best solution for everything, which isn't really always the case. In > this case, you've got two processes running on an OS with pretty poor > process management and a networking architecture that doesn't appear to > be optimized for communication between local apps. Those are two things > that are much improved on Mac OS X, fortunately. > > Though I don't have much experience with the CodeWarrior debugger -- > println's and the mrj dcmd can actually accomplish quite a lot if you > need them to :) -- I suspect that remote debugging may perform > reasonably close to local debugging, if not better, since you don't have > the overhead of the process switching. Give that a shot if you can set > two machines up next to each other. > > -Eric
Re: Mac Java debugging
Author: Eric Albert
Date: Sun, 22 Oct 2000 00:00
Date: Sun, 22 Oct 2000 00:00
20 lines
1261 bytes
1261 bytes
In article <221020001305457266%anon@nowhere.com>, Anon Anonymous <anon@nowhere.com> wrote: > I only used C++ under CW 5.3. Now that I am working on Java 1.1 > stationary projects on the Mac with CW Pro 6.0, I can see the problem > people are experiencing. > > If the speed issue is new to CW 6, and did not exist in 5.3, then I > would think the problem lies with CW 6, and not Apple, Mac OS > limitaitons, etc. > > Are ther any 5.3 users having problems with slow java debuggers? Yep. It's been a very longstanding complaint. :( -Eric -- Eric Albert ejalbert@stanford.edu http://www.stanford.edu/~ejalbert/
Thread Navigation
This is a paginated view of messages in the thread with full content displayed inline.
Messages are displayed in chronological order, with the original post highlighted in green.
Use pagination controls to navigate through all messages in large threads.
Back to All Threads