This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
Please see the discussion at NBDEV: http://www.netbeans.org/servlets/ReadMsg?msgId=390267&listName=nbdev I will attach a patch that begins the work. It is not complete; in particular, I have not done the update for the website, I haven't finalized the work on the Bundle.properties, etc. I will need help from within Sun making sure that <scramble> and <unscramble> work correctly. Also, somebody needs to take care of the licensing stuff (discussed extensively in the mailing list thread above). Currently, the patch is against ANTLR 2.7.2a3. To try it out, you will need to download: http://www.antlr.org/download/antlr-2.7.2a3_9-14-2002.tar.gz Inside the tar.gz is a file antlrall.jar. Copy the file antlrall.jar to ${nb_all.dir}/libs/external/antlrall-2.7.2a3.jar This patch creates a single module that provides access to the classes necessary to support language processors that were created using the ANTLR compiler compiler (the ANTLR runtime library). It also provides access to all of the classes that are needed by tools like the ANTLR module (http://antlr.netbeans.org ) to support building ANTLR grammars. If so desired, I can split it up into two modules (the ANTLR runtime is ~150KB uncompressed, the ANTLR development tool classes are ~450KB uncompressed).
Created attachment 7679 [details] Patch for existing files in libs/
Created attachment 7680 [details] New files (extract inside libs to create libs/antlr)
Brian for licensing issues please contact lenka.kasparova@sun.com. If I remember well, Mila wanted ANTLR runtime for lexer module. So you could share your effort.
A new version of ANTLR is going to be released sometime soon. The current version is in the public domain but the maintainers are considering using a BSD license for the next release. So, I'm waiting for the final release before preceding.
ANTLR 2.7.2 has been released. It is still in the public domain. Previous versions of ANTLR were split into two parts: the runtime library and the development library. The runtime library contained only the classfiles necessary for an ANTLR-based application to run properly. The development library included the actual ANTLR-to- Java/C#/C++ compilers. I was intending to create two modules, one corresponding to each ANTLR library. However, the ANTLR 2.7.2 build system changed and now everything gets packaged together into a single large JAR file. Furthermore, the JAR file is compressed. Here are the file sizes (runtime-only sizes are from ANTLR 2.7.0) combined runtime-only compressed 360K 60K uncompressed 685K 108K Is the 360K compressed, combined JAR file acceptable? Or, should I repackage the ANTLR classes into two uncompressed files (108K antlr-runtime.jar and ~550K antlr-dev.jar)? My understanding is that NetBeans policy is to avoid compressed JAR files. I believe that most modules that use ANTLR only need the 108k runtime system; only the ANTLR module itself (antlr.netbeans.org) would need the development system. So, I am leaning toward the second option. But, that option requires a NetBeans-specific build script for ANTLR.
I would also opt for the second option but I'm not sure what the consequences are. Do they plan to silently intermix the runtime classes with the design-time classes at a functional level? I've looked to antlr.org but didn't find any clue. I'm wondering whether we will be able to cleanly determine into which of the two jars a particular class will go. I mean the compile-time jar for the previous versions of antlr was easy to generate as the runtime jar was provided in the distribution so it only required to take all classes under antlr dir and remove those contained in the runtime jar already. The remaining classes formed the compile-time jar. I'm worried that it will no longer be so easy. Regarding compression of jar files I think we should be OK even with the compressed jar files. There was nbdev thread about it http://www.netbeans.org/servlets/ReadMsg?msgId=385637&listName=nbdev Added Jesse to cc if he has any comments on this.
Re. compression? No idea, ask Petr.
No strong opinion on it. I'd use split uncompressed jars, but for hard numbers, I'd have to do some measurements on different machines ({slow,fast}x{CPU,disk})
Brian have you applied for legal review at nbdiscuss? If you have LFI please refer it in issue status whiteboard in form LFI=0000.