DRLVM does not print anything - launcher does not pass -X to DLRVM.
Out of date information - Euege Ostrovsky has fixed this already :) Ok, looks like WEB page & URL from 'java -X' is a solution. Let's follow this one. Thanks for you feedback. Vladimir Beliaev 2007/6/20, Pavel Pervov <[EMAIL PROTECTED]>:
Why adding URL to -X? Help on -X should be printed with 'java -X'. BTW, DRLVM does not print anything - launcher does not pass -X to DLRVM. Pavel. P.S. BTW, what I've proposed will solve this issue. On 6/20/07, Gregory Shimansky <[EMAIL PROTECTED]> wrote: > > Pavel Pervov wrote: > > Vladimir, > > > > 'java -X' usually prints out help on options, which start with -X. > > > > What you was asking about as I understood that has syntax of > > -XX:<vm_option>[=<value>]. Sun's VM has help on these options on the Web > > page and no means to print it out to console. > > > > So, I'm -1 here for printing help on -XX options. > > I agree that it is better to provide help on DRLVM internal properties > on web site instead of printing it to console. We can also have a man > page in the future which would include that text too. > > Adding an URL to help on -X is a good idea. > > > If community still decide to print such help - there is 4-th way to do > so > > (which will require changing.launcher). > > 1) launcher recognizes several generic options and is able to determine > > application class or jar name. > > 2) if no -vmdir or -vmdll is specified on command line - launcher tries > to > > load that VM, otherwise defaults to <launcher > location>/default/harmonyvm > > and tryies to load this one. If it fails - it prints short help message > as > > it does right now, when launched with some unrecognizable property name. > > 3) if library is loaded successfully - launcher constructs arguments > array > > (as it does right now) and calls to JNI_CreateJavaVM from loaded VM > > library. > > 4) if this call returns (-?, -h, -help and -X do not return from > > JNI_CreateJavaVM) and if launcher finds class of jar name on the command > > line - it proceeds with program startup. > > > > What do you think on such generic startup sequence? > > > > WBR, > > Pavel. > > > > On 6/20/07, Vladimir Beliaev <[EMAIL PROTECTED]> wrote: > >> > >> Hello, All > >> > >> I want to add a *help *for drlvm specific system properties like '- > >> Xvm.assert_dialog=false' (as asked in > >> HARMONY-3409<http://issues.apache.org/jira/browse/HARMONY-3409> > >> ). > >> > >> *Could you send your opinion on what way this can be done?* There are > >> three > >> safe ways to do this & non of them seems to be really good (please see > >> technical details in > >> HARMONY-3409<http://issues.apache.org/jira/browse/HARMONY-3409>). > >> > >> > >> *way 1*: support 'java -Xhelp:prop foo' command line... Like if one > runs > >> 'java -X' then this help says "run '-Xhelp:prop foo' to see sytem > >> properties"... In short: if one does not specify this 'foo' at the end, > >> then > >> a generic launcher consider command line as broken (no 'mainClass' is > >> specified) & does not invoke JNI_CreateJavaVM (which print a help)... > >> > >> *way 2*: support 'java -Xhelp:prop' command line (i.e. w/o dummy > >> 'foo')... > >> To implement this the *generic* launcher is to be changed to recognize > >> '-Xhelp:' as a 'help' option (like it recognizes ? -h -help -X now). > >> Because > >> of other VMs do not support such a '-Xhelp:' option we get > >> *generic*launcher to be a bit drlvm-specific... > >> > >> *way 3*: be like other VMs and do not print a 'help for drlvm internal > >> properties'. Instead, 'java -X' help output may provide a URL to > >> harmony.apache.org html document explaining drlvm internal system > >> properties... > >> > >> I would pick a *way 1* as it does not change *generic *launcher & we > are > >> free to do every thing we want with -X... I do not like *way 3* at all > - > >> it > >> is not a big deal to create such a document, still it is one more > >> document > >> to be supported (I believe it may become out-of-date pretty in couple > >> months)... > >> > >> So, what do you think? > >> > >> -- > >> Thanks > >> Vladimir Beliaev > >> Intel Middleware Products Division > >> > > > > > > > > > -- > Gregory > > -- Pavel Pervov, Intel Enterprise Solutions Software Division
