🚀 go-pugleaf

RetroBBS NetNews Server

Inspired by RockSolid Light RIP Retro Guy

Article View: comp.sys.mac.programmer.codewarrior
Article #105

Remote java debugging very slow from OS X

#105
From: Mike Fahl
Date: Thu, 07 Feb 2002 14:36
47 lines
3377 bytes
I'm looking at migrating my java development to CodeWarrior 7.1 under
OS X. Most of the time I'm cross debugging to a PC. However, JDWP
cross debugging from an OS X G4/500 development host to a 500 MHz
Win2k box is *extremely* slow:

* Launching the app under remote debugging takes 30 seconds, timed
from hitting Debug in CW to the first window appears on the screen
(this is a moderately sized Swing app).

* Continuing from there until I hit the first breakpoint (set to fire
when I open a dialog box from within my app) takes 19 seconds the
first time (drops down to 2-3 seconds for subsequent hits of the same bp).

* Single stepping takes about one second per step.

It doesn't seem that the speed of the debug target makes much
difference. During these long pauses I see a furious flashing on my
hub, as if many small packets and handshaking takes place between the
OS X Mac and the debugged PC (this is all within one room, with a 100
MBit hub and cables known to be good). My guess is that the VM being
debugged passes a lot of data back to CW about classes being loaded,
etc, since this is mainly during app start-up and the first time I
access classes (eg, open a dialog box).

Now comes the fun part. Doing the above but running CodeWarrior 7.1
from a PC WinNT host instead (with the same Win2k remote debug target)
gives the following times:

* Launching the app under remote debugging takes 4 seconds.

* Continuing from there until I hit the first breakpoint takes 1
second the first time.

* Single stepping is virtually instantaneous.

During those initial four seconds, I see a relatively small number of
flashes on the hub, as if the number of packets is much smaller (but
possibly larger - I haven't sniffed them).

My conclusions: Either the TCP/IP implementation in OS X sucks, or
there's a bug/deficiency in CW7.1 for OS X. Could it be that CW7 goes
through the OpenTransport shim rather than straight to the sockets
implementation, and perhaps that's slowing things down? But this badly?

Any ideas or suggestions are most welcome!

Mike Fahl
Technical Details
Message-ID: <3C6290C1.DA403898@dataton.se>
Path: arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!newsfeeds.belnet.be!news.belnet.be!news.tele.dk!small.news.tele.dk!194.213.69.151!news.algonet.se!algonet!news2.tninet.se!not-for-mail

Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit
X-Google-Thread: 10bcce,4609b529324afc4c,start
X-Google-Attributes: gid10bcce,public
X-Google-Thread: 111f0d,4609b529324afc4c,start
X-Google-Attributes: gid111f0d,public
X-Google-Thread: f635e,4609b529324afc4c,start
X-Google-Attributes: gidf635e,public
X-Google-ArrivalTime: 2002-02-07 07:33:32 PST
Path: arc1-ca.pugleaf.net!archiver1.google.com!news1.google.com!newsfeed.stanford.edu!newsfeeds.belnet.be!news.belnet.be!news.tele.dk!small.news.tele.dk!194.213.69.151!news.algonet.se!algonet!news2.tninet.se!not-for-mail
From: Mike Fahl <fahl@dataton.se>
Newsgroups: comp.sys.mac.programmer.codewarrior,codewarrior.java,codewarrior.mac
Subject: Remote java debugging very slow from OS X
Date: Thu, 07 Feb 2002 16:36:02 +0200
Organization: Dataton - TRUE MULTIMEDIA
Lines: 47
Message-ID: <3C6290C1.DA403898@dataton.se>
Reply-To: fahl@dataton.se
NNTP-Posting-Host: 195.163.32.202
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: green.tninet.se 1013096029 12358 195.163.32.202 (7 Feb 2002 15:33:49 GMT)
X-Complaints-To: postmaster@telenordia.se
NNTP-Posting-Date: Thu, 7 Feb 2002 15:33:49 +0000 (UTC)
X-Mailer: Mozilla 4.76 (Macintosh; U; PPC)
X-Accept-Language: en