review request for 6798511/6860431: Include functionality of Surrogate in Character

Ulf Zibis Ulf.Zibis at gmx.de
Thu Aug 27 20:59:02 UTC 2009


Am 27.08.2009 22:17, Xueming Shen schrieb:
> I'm reading all those 100xxx one by one. The problem is that you've 
> mixed too many things together in one bag and there are
> too many "dependencies" among them.

Yes, there are indeed many dependencies in 100098. The before 100098 
changes have been more simple, but unfortunately they are mostly 
obsolete in face of 100098.

> There are lots of good idea/suggested changes,

Thanks for the flowers :-)

> but it really takes time to figure out which
> one is the real goal of one particular patch,  which one has the 
> priority, and how to "extract" something out of hundreds of changes
> to do it step by step.

I can understand this. I tried to keep the amount "small", but after 
some time, I saw, that I had no idea how (OK, there are some 
supplementary cosmetics).
I "only" made the changes to a part of all charsets. As result I had to 
remain 3 different patterns for instantiation of charsets in 
ExternalCharsetProvider.
The changes to the UTF classes should be good candidate for extraction.

> I can't not just take such huge changeset and throw into the JDK.

That's reasonable. Should I make future changesets on base of the one 
before e.g. 100098 instead on base of tl forest?

-Ulf





More information about the core-libs-dev mailing list