[x3d-public] X3DJSAIL/X3DJSONLD Status report: 4% Java files that do not compile. 91% runtime JSON output (not = identical JSON yet)

yottzumm at gmail.com yottzumm at gmail.com
Thu Mar 23 22:27:24 PDT 2017


Of the 394, 315 don’t compile, with “code too large” errors.  This means the method is too large I think.  22 of those have two many constants, so I think we could compile around 293 more files with some effort in the serializer breaking up the method.  So now with bug fixes, I have 342 files that won’t compile, and 25 that might be able to be fixed easily with some work.  Note, I have reduced the number of compiler errors to 800, but I think I was undercounting before (I wasn’t counting the files with 1 error). 20 compilations have more than 2 errors and have 449 errors total.  I have a zip with the 20 files here + the errors.  http://coderextreme.net/Java3Errors.zip.  Don, could you look at these and see if you can make API fixes for any of them?  Thanks!  Many of them are still script errors which I haven’t fixed, so ignore those.  I am thinking F16.java and the ones with .set*().  I think I may have a handle on MultiTextureDesignPattern.java.

Looks like after fixing the serializer and scripts, we are almost ready to move to runtime errors, if Don isn’t already there (are you working on the runtime errors I have been reporting, the ROUTEs and USE container field errors?)

And we’ll also address the files with 1 and 2 errors which don’t come out with “code too large”

John
Sent from Mail for Windows 10

From: yottzumm at gmail.com
Sent: Thursday, March 23, 2017 1:05 AM
To: Don Brutzman; X3D Graphics public mailing list; Roy Walmsley
Subject: X3DJSAIL/X3DJSONLD Status report: 4% Java files that do not compile. 91% runtime JSON output (not = identical JSON yet)

With older versions of X3DJSAIL,

Of the 15313 Java files in my testing collection, 394 do not compile.  The rest compile or have compiled in the past.  That’s 96% of the files compile.  This does not include files which don’t match the Cp1252 encoding.  I have deleted those from my testing collection.
That’s out of  15313 JSON files.  Some of these have JSON schema issues or bad URLs.  I think that number is 331, so there must be some Java files that don’t come from JSON.

I will report on the Java compiler errors next.  My initial estimate for the number of errors was 345, which is way below the number of files, so I am thinking perhaps there could be more files which compile, and they were prevented by compilation errors.  I am now compiling the files one at a time (actually 8 at a time with “xargs -L 1 -P 8”).

Once I regenerated the files and recompiled the ones that didn’t compile before, , I got 507 (or 394, more likely, not including existing Java files from X3D resources, which may have classpath issues.) which do not compile, and a total of 2323 outstanding errors, 4-5 per file on average.  That’s 96% of the Java files compiling at one point.  I still need more work on my serializer, obviously!

I probably need to delete the .class files and start over, but I’m having fun with a bit of statistics!

In other news, 14009 of the class files produced result output JSON files. That’s pretty good! 91%  Effort on previously reported X3DJSAIL runtime errors is appreciated!  I will now look at the new version!  To do:  Diff original JSON with new JSON.

This is out of a total of 15651 .x3d input files.

Java that doesn’t compile is here: http://coderextreme.net/NonCompiledJava.zip if someone wants to help with the remaining 394 Java files.  This archive may disappear (40MB) as it is large.

It would be nice if we could move more of the errors from runtime to compile-time issues if possible (4% compile time and 5% runtime errors), can we do more work on OM4X3D or X3D or X3DJSAIL to make them compile-time?  This should be investigated.  It may be that my use of the library is more apt to make them runtime issues.  Contributions to my JavaSerializerNew.js and associated files are welcome!

John

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://web3d.org/pipermail/x3d-public_web3d.org/attachments/20170324/13c20501/attachment.html>


More information about the x3d-public mailing list