[asdf-devel] Pushed proposed fix for Xach's bug

[Robert Goldman]

On 9/2/11 Sep 2 -5:41 PM, Faré wrote:
> On Fri, Sep 2, 2011 at 18:29, Robert Goldman <rpgoldman at sift.info> wrote:
>> Passes all of the unit tests (including the new one based on Xach's test
>> case).
>>
>> Please, all of you, have a peek at the diffs (possibly I should have
>> squashed the changeset to make this easier, sorry), and the commit
>> message, and test it as soon as possible.
>>
>> I have not bumped the version number, as this is still a work in
>> progress subject to review.
>>
> If you push a code change to master, you MUST bump the version number.
> If you don't feel confident about the code, please don't push to
> master.

OK, sorry about that.  I did the bin/bump-version, and pushed.

As for pushing to master, while I am not perfectly confident in the
patch, it passes all the tests, and I'm as confident as I can get
without people on the bleeding edge giving it a test.

> Didn't you say it would be better to throw the old object and create a
> new one than to try to reinitialize those component instances?

Yes, but Christophe convinced me that this was The Wrong Thing, because
methods defined on (EQL (FIND-SYSTEM <foo>)) would be broken by doing
so.  That led me to the REINITIALIZE-INSTANCE approach.  Note, though,
that it should have the same semantics --- the COMPONENTs are cleared,
and that means that the children of the system will be made new.  So I
believe only the SYSTEM objects will be reused, since once the children
slot of the SYSTEM is cleared, the previous child COMPONENTs should
become unreachable.

I understand you are busy with other things, but if you get a chance,
LMK what you think.

Best,
R