[vmkit-commits] [PATCH] Remove System.err.println from mmtk stuff to fix dep on bootstrap

[Nicolas Geoffray]

Hi Will,

On Thu, Nov 3, 2011 at 3:48 PM, Will Dietz <wdietz2 at illinois.edu> wrote:

> Inlined below.
>
> This change seems benign, but I'm not sure if it's masking some other
> bug I should be concerned about.
>
> Patch primarily a discussion point, although could be committed too
> (since it's rather minor).
>
> FWIW, with this patch (and the previous ones), we can build a
> OpenJDK-based VMKit that can run many simple codes, and on larger
> codes seems to only fail when encountering some of the many NYI()'d
> (or intentionally unimplemented) pieces.  Which is kinda neat! :)
>

Definitely! That's a giant step! Any plans on sending the instructions on
which OpenJDK to use? How to configure, etc? :)


> In the future I hope to focus on getting it to run various codes
> better (dacapo being my big goal, but not the full story), and try to
> generally improve misc behavior compatibility/compliance-wise :).
>

Big encouragements for getting dacapo to run, especially all benchmarks of
the latest version (which do not run on GNU Classpath).

There is also SpecJVM2008 I think, that anyone can download. It'd be neat
to have that running too :)


>
> Thanks!
>
> ~Will
>
> >From 0462e91eb86e28c00707c40d624853eb51679904 Mon Sep 17 00:00:00 2001
> From: Will Dietz <w at wdtz.org>
> Date: Wed, 2 Nov 2011 23:02:45 -0500
> Subject: [PATCH 15/17] Remove System.err.println from mmtk stuff to fix dep
>  on bootstrap
>
> Not entirely sure, but with these present we get linker errors with
> libFinalMMtk.a on misc pieces that are pulled in from System.err,
>  namely things like sun.nio.cs.StreamEncoder.
>

This may just be a limitation of the AOT compiler, which pulls out more
than it actually needs. I'd be interested to see the error messages if you
still have them.

Patch is fine to commit because it just touches mmtk/src/org/j3 files. Note
that if it were to touch other mmtk/src/org files, it wouldn't be as fine
because it's code that is imported from mmtk. And this code we should try
to not deviate from it, to avoid merge conflicts.

Again, that's a great step forward! Nice work. I'm looking forward to see
all the dacapo benchmarks run on vmkit :)

Nicolas


> On one hand this seems like an undiagnosed failure of the precompilation,
>   however we're not trying to build System.err things while
>   compiling mmtk, and this change is relatively minor.
> ---
>  mmtk/java/src/org/j3/mmtk/BuildTimeConfig.java |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/mmtk/java/src/org/j3/mmtk/BuildTimeConfig.java
> b/mmtk/java/src/org/j3/mmtk/BuildTimeConfig.java
> index 8dc8f82..9f90b8e 100644
> --- a/mmtk/java/src/org/j3/mmtk/BuildTimeConfig.java
> +++ b/mmtk/java/src/org/j3/mmtk/BuildTimeConfig.java
> @@ -53,7 +53,7 @@ public final class BuildTimeConfig extends
> org.mmtk.vm.BuildTimeConfig {
>     if (default_property_file == null) {
>       propFileName = System.getProperty(property_file_property);
>       if (propFileName == null) {
> -        System.err.println(property_file_property+" must specify a
> properties file");
> +        //System.err.println(property_file_property+" must specify a
> properties file");
>         VM.sysExit(1);
>       }
>     } else {
> @@ -67,7 +67,7 @@ public final class BuildTimeConfig extends
> org.mmtk.vm.BuildTimeConfig {
>       propFileStream.close();
>     } catch (FileNotFoundException e) {
>       if (!propFileName.equals(default_property_file)) {
> -        System.err.println(propFileName+" not found.");
> +        //System.err.println(propFileName+" not found.");
>         VM.sysExit(1);
>       }
>     } catch (IOException e) {
> --
> 1.7.5.1
> _______________________________________________
> vmkit-commits mailing list
> vmkit-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/vmkit-commits
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/vmkit-commits/attachments/20111104/be7d5faf/attachment.html>