[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [aspectj-dev] incremental support on the command-line
|
Just feedback from the developers in the trenches: I'm really glad to hear that the incremental support is getting more attention and that these use cases will be tested.
Cheers,
Nick
P.S. While on the topic, I've noticed that incremental compilation deals rather poorly with static inner aspects. A full recompile is often needed to make them work. That's a pity when working in AJDT. I know I should write a bug/test case for it, but haven't isolated it in a reproducible form.
On Mar 16, 2004, at 1:09 AM, Adrian Colyer wrote:
I'm not sure of the historical reasons for that limitation (expediency of implementation?), but I can't immediately see any reason why we couldn't support incremental compilation using a list of source files as opposed to sourceroots (by the time we get to actually compile, the origin of the source files is lost anyway). A full ".lst" file has more implications - it can contain a whole host of compiler directives. We also don't say in the docs (that I could find) whether we support injars, inpath and aspectpath with incremental compilation. They currently don't work (any changes in resources on these paths will be ignored I believe), but the compiler accepts them. Copying of resources (only an issue if inxxx is allowed) is also broken under incremental (no changes detected again). We also allow the user to specify -outjar, but that's broken too.
[Just to qualify - when I say 'broken', I don't mean it all falls down in a heap, just that not all cases are handled correctly.]
All of these things will be needed by AJDT, and I propose to extend our incremental capability to cover them (even if in the very simplest case we simply revert back to a batch build on change detection). Once that is done and the tests are in place, I see no reason to continue to limit the command line (or ant) incremental compiles. AJDT has to cope with classpath etc. changes between compiles, and will follow the JDT rule - if any path element changes, then do a full build.
I raised bugs 54618, 54621, 54622, and 54625 last week to cover these issues. The next challenge is supporting tests of this nature in the test harness...
-- Adrian
Adrian_Colyer@xxxxxxxxxx
<x-tad-smaller>Wes Isberg <wes@xxxxxxxxxxxxxx></x-tad-smaller>
<x-tad-smaller>Sent by: aspectj-dev-admin@xxxxxxxxxxx</x-tad-smaller>
<x-tad-smaller>16/03/2004 01:40</x-tad-smaller>
<x-tad-smaller>Please respond to aspectj-dev</x-tad-smaller>
<x-tad-smaller> </x-tad-smaller>
<x-tad-smaller> To: "aspectj-dev@xxxxxxxxxxx" <aspectj-dev@xxxxxxxxxxx></x-tad-smaller>
<x-tad-smaller> cc: </x-tad-smaller>
<x-tad-smaller> Subject: [aspectj-dev] incremental support on the command-line</x-tad-smaller>
<x-tad-smaller>We used to say -sourceroots is required to support incremental on </x-tad-smaller>
<x-tad-smaller> the command-line, but we seem to be able to implement incremental</x-tad-smaller>
<x-tad-smaller> using .lst files in AJDT. True? Will there be a continued</x-tad-smaller>
<x-tad-smaller> need for this limitation on the command-line compiler in 1.2?</x-tad-smaller>
<x-tad-smaller> Thanks -</x-tad-smaller>
<x-tad-smaller> Wes</x-tad-smaller>
<x-tad-smaller> _______________________________________________</x-tad-smaller>
<x-tad-smaller> aspectj-dev mailing list</x-tad-smaller>
<x-tad-smaller> aspectj-dev@xxxxxxxxxxx</x-tad-smaller>
<x-tad-smaller> http://dev.eclipse.org/mailman/listinfo/aspectj-dev</x-tad-smaller>
