Re: Tya vs. shuJIT
David Brownell wrote: > It'd be interesting to see a current version of GCJ in those > comparisons. GCJ 2.95.2 is listed in the shudo.net page, > but that's _extremely_ old ... I'd suggest using the 2.96 that > is distributed with RedHat 7.0, as the most current "stable" > version available. The newest stable version of GCC is still 2.95.2. Please see http://www.assurdo.com/dd.sh/web-server/ This page says, GCC 2.96 is not a formal GCC release. GCJ has been changed from 2.95.2 very much. But, as you said, the current codebase seems to be unstable. I don't wonder even if it cannot be compiled. I'm also very interested in evaluating newer GCJ, but I don't want to be bothered with such problems. If there is a binary package (i.e. RPM) of a snapshot, I'd like to measure its performance. > (GCC 3.0 will have a more up-to-date > version, although still without serious optimizations. But > that codebase is still slushing, in prep for a release.) Kazuyuki SHUDO Happy Hacking! Muraoka Laboratory, Waseda University -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: server sees no body when doing HTTP POST
> "John D. Mitchell" <[EMAIL PROTECTED]>: > In general, with respect to dealing with HTTP in Java, rather than > reading all of the related RFCs and trying to implement them > yourself, Hm... I'm not sure if I get your meaning here...? Using URLConnection is definitely a step up from implementing HTTP over naked sockets. Or maybe you're referring to my manual construction of a multipart/form-data body? Being able to remove that code, would definitely be a good thing. > I strongly suggest that people use (and enhance!) an existing HTTP > client package such as the open source HTTPClient, > http://www.innovation.ch/java/HTTPClient/ I've encountered from time to time in www.google.com searches, but I haven't given it a closer look. I have preferred to stay within the class libraries offered by JDK. One reason to give it a try, is that HttpOutputStream seems to be able to write data directly to the network, if the other end is HTTP 1.1 capable. Right now I'm getting java.lang.OutOfMemory exceptions, when trying to write a 12MB file in an HTTP POST: java.lang.OutOfMemoryError: at java.io.ByteArrayOutputStream.write(ByteArrayOutputStream.java) at java.io.OutputStreamWriter.flushBuffer(OutputStreamWriter.java) at java.io.OutputStreamWriter.flush(OutputStreamWriter.java) at java.io.BufferedWriter.flush(BufferedWriter.java) at java.io.PrintWriter.newLine(PrintWriter.java) at java.io.PrintWriter.println(PrintWriter.java) at RepositoryCmd.send(RepositoryCmd.java:72) at Upload.main(Upload.java:32) I'm guessing that the culprit here, is the ByteArrayOutputStream. However, why it should crash at a mere 12MB is beyond me. According to top it gives up way before it can make a serious impact on the available swap space? Is there some command line argument I have to give to the java runtime, I wonder? -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Adjusting heap size (Was: server sees no body...)
> Steinar Bang <[EMAIL PROTECTED]>: > Right now I'm getting java.lang.OutOfMemory exceptions, when trying to > write a 12MB file in an HTTP POST: > java.lang.OutOfMemoryError: > at java.io.ByteArrayOutputStream.write(ByteArrayOutputStream.java) > at java.io.OutputStreamWriter.flushBuffer(OutputStreamWriter.java) > at java.io.OutputStreamWriter.flush(OutputStreamWriter.java) > at java.io.BufferedWriter.flush(BufferedWriter.java) > at java.io.PrintWriter.newLine(PrintWriter.java) > at java.io.PrintWriter.println(PrintWriter.java) > at RepositoryCmd.send(RepositoryCmd.java:72) > at Upload.main(Upload.java:32) > I'm guessing that the culprit here, is the ByteArrayOutputStream. > However, why it should crash at a mere 12MB is beyond me. According > to top it gives up way before it can make a serious impact on the > available swap space? > Is there some command line argument I have to give to the java > runtime, I wonder? Yes, there is. "-mx" sets the maximum heap size. This is 16MB or thereabouts by default. Setting it to 2400 and to 4000 didn't help, but setting it to 8000 did. I guess reallocating the byte array when neccessary, requires a bit of space? Hm... maybe I can set the size of the byte array before writing into the ByteArrayOutputStream? I know how large it at least has to be from the size of the file I'm trying to POST. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Tya vs. shuJIT
Sorry for my mistake on an URI. > The newest stable version of GCC is still 2.95.2. > Please see http://www.assurdo.com/dd.sh/web-server/ > This page says, GCC 2.96 is not a formal GCC release. The above URI should be: http://gcc.gnu.org/gcc-2.96.html The original false URI points the web page of a small web server implemented with /bin/sh and /bin/dd. It has no relation to Java/Linux at all. Kazuyuki SHUDO Happy Hacking! Muraoka Laboratory, Waseda University -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Tya vs. shuJIT
> From: "Kazuyuki Shudo" <[EMAIL PROTECTED]> > Sent: Tuesday, January 30, 2001 12:23 AM > > David Brownell wrote: > > > It'd be interesting to see a current version of GCJ in those > > comparisons. GCJ 2.95.2 is listed in the shudo.net page, > > but that's _extremely_ old ... I'd suggest using the 2.96 that > > is distributed with RedHat 7.0, as the most current "stable" > > version available. > > The newest stable version of GCC is still 2.95.2. > Please see [ http://gcc.gnu.org/ ] > This page says, GCC 2.96 is not a formal GCC release. That's why I put "stable" in quotes ... and referred to GCJ, not GCC! And specified RedHat. The GCC maintainers are finally, on the eve of 3.0, talking 2.95.3 ... but that's not useful if you are interested in a Java compiler. I suspect they've been focussed mostly on the traditional C core of the GNU Compiler Collection, not the C++ or Java parts. > If there is a binary package (i.e. RPM) of a snapshot, > I'd like to measure its performance. It was recently pointed out on the GCJ list that the RedHat RPM is the best widely available version of GCJ. I know that I was pleased not to have the headache of trying to build it without the (now) integrated GCC/libgcj trees. For that, I'm more than willing to use "kgcc" for kernels! - Dave > > (GCC 3.0 will have a more up-to-date > > version, although still without serious optimizations. But > > that codebase is still slushing, in prep for a release.) > > Kazuyuki SHUDO Happy Hacking! > Muraoka Laboratory, Waseda University -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
writing data to lp0
Hi, I want to write some data to my lp0-port. Does anyone know how I can do this? Thanks, Raphael -- E-Mail: [EMAIL PROTECTED] - www: http://www.rampro.de - icq: 59473818 Linux is like a wigwam: no windows, no gates and apache inside. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Does the -O option affect speed?
I was wondering if compiling with the -O options has any significant effect on speed. The jikes '-help' message indicates that -O suppresses the line number table. That would help save memory, but does it affect speed? Does it depend on the jvm being used? (I use Blackdown 1.3 with HotSpot turned on.) Thanks, Barnet Wagman -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
