[rfc][icedtea-web] Makefile New Whitelisted-Dist-Test option

[Jie Kang]

----- Original Message -----
> On 09/22/2014 11:17 PM, Jie Kang wrote:
> > Hello,
> >
> > This patch expands on the Makefile Reproducer Test patch [1].
> >
> > [1]
> > http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2014-September/029581.html
> >
> >
> > There is now a new make rule: 'run-netx-whitelisted-dist-tests'
> >
> > This rule runs the reproducers similar to 'run-netx-dist-tests', except it
> > only processes (compiles, etc.) the resources filtered by the whitelist.
> >
> > As well, the rule 'run-netx-dist-tests' has now reverted back to processing
> > all resources (ie. what it used to do before the Makefile Reproducer Test
> > patch [1])
> >
> > Thoughts?
> >
> > Just a note, the purpose of the "stamps/whitelist-filter.stamp" rule
> > having:
> >    echo ".*" ...
> > is to maintain compatibility. Though the "all-whitelist" rule also contains
> > duplicate code, there can be situations where neither "all-whitelist" or
> > "filtered-whitelist" are not called. E.g. if user runs "make
> > stamps/netx-dist-tests-prepare-reproducers.stamp". This is the solution I
> > came up with to allow for all possible make commands to continue to work
> > and for the user to be able to quickly switch between running "make
> > run-netx-dist-tests" and "make run-netx-whitelisted-dist-tests" without
> > always having to perform "make clean-netx-dist-tests". I guess a TLDR is
> > that it's to prevent regressions.
> >
> > If this patch is accepted I will update the wiki and documentation for this
> > feature.
> >
> >
> > Regards,
> >
> 
> Hm. only hint to approach. Why  not configure switch? This seems to me
> overcomplexed.  What is
> advantage of this?

Hello,

I have attached a completely new patch that follows your hint. Thanks a lot, it is much better now, and more simple. 
New option for configure: --enable-whitelist-processing

With the flag on, the Makefile generated will filter by whitelist for processing, otherwise, it will process all, just like before patch [1].

[1] http://mail.openjdk.java.net/pipermail/distro-pkg-dev/2014-September/029581.html

> Once the new and old targets will be run in unpredicted order, the result can
> be unpredictible.

The complexity was to make sure running new and old targets in random order still worked hahah. And you could run them and stop them whenever, and still work properly like before too! But now no longer needed so...

> Also I can not see the reverted part in patch.

There wasn't really any reversion in code, just reversion in behaviour. Anyways, no longer relevant.


Thanks a lot!!


Regards,

> 
> J.
> 

-- 

Jie Kang
-------------- next part --------------
A non-text attachment was scrubbed...
Name: itw-make-configure-whitelist-1.patch
Type: text/x-patch
Size: 2377 bytes
Desc: not available
URL: <http://mail.openjdk.java.net/pipermail/distro-pkg-dev/attachments/20140923/31706462/itw-make-configure-whitelist-1.patch>