[foreign-jextract] RFR: 8253725: Jextract fails to extract big monolithic headers [v2]

[Jorn Vernee]

On Mon, 28 Sep 2020 17:43:06 GMT, Maurizio Cimadamore <mcimadamore at openjdk.org> wrote:

>> This patch adds support for splitting the symbols in the main header class into multiple classes; the header class is
>> then rewrited to extend all these additional classes, so that all symbols would be exposed to users in the classic way.
>> I've toyed with some ways to split on a by-header basis, but found that in certain cases we would generate near-empty
>> classes - and that would still not prevent issues from occurring in the corner case a single very big header file
>> contained so many symbols.  I've reorganized the code a bit - I noted that we were generating several nested classes,
>> but in some cases (structs) we were using a builder, in other cases (functional interfaces, typedef) we weren't. Now we
>> use builders for classes, and I've added an intermediate nested builder abstraction which does the right thing for all
>> nested cases.  I've also converted the source constant generator to be a JavaSourceBuilder too - since it too is a
>> class, after all, with a classBegin/End and build methods.  Finally, I moved the string builder/indentation logic out
>> to a different class (StringSourceBuilder); since some builders want to have their own string buffer, while others (the
>> nested ones) don't.  The organization of the builder code is still far from optimal, but I think that, moving forward,
>> this consolidation should make it easier to make changes.
>
> Maurizio Cimadamore has updated the pull request incrementally with one additional commit since the last revision:
> 
>   Remove redundant import generation

Marked as reviewed by jvernee (Committer).

-------------

PR: https://git.openjdk.java.net/panama-foreign/pull/357