[swift-users] Annotating C APIs without changing the original header files

[Daniel Dunbar]

We don't have explicit support for api notes in SwiftPM.

We discussed it, and it something which probably makes sense, but no one has worked on a design or implementation yet.

 - Daniel

> On May 12, 2017, at 11:32 AM, Michael Gottesman via swift-users <swift-users at swift.org> wrote:
> 
> +Ankit
> 
> Michael
> 
>> On May 12, 2017, at 10:10 AM, Geordie J via swift-users <swift-users at swift.org <mailto:swift-users at swift.org>> wrote:
>> 
>> To continue this thread: I managed to annotate a bunch of C APIs with modulename.apinotes. This works with Xcode (to a certain degree - pointers, enums, and especially OpaquePointers are tricky). I’m now trying to build my package with SwiftPM and it doesn’t seem to recognise the apinotes file. 
>> 
>> @Doug Gregor, would you be able to advise as to whether apinotes works with SwiftPM (on Linux) and whether it requires some extra settings that I may be unaware of?
>> 
>> Thanks and best regards for the weekend,
>> Geordie
>> 
>> 
>>> Am 08.05.2017 um 00:51 schrieb Geordie Jay <geojay at gmail.com <mailto:geojay at gmail.com>>:
>>> 
>>> I'm having the same issue. The renames seem to work, as in they disappear from the global scope with a fixit to rename to the new (namespaced) version if I type in the name manually, but they don't appear as static members of the enum type, regardless of how I call them. Would appreciate some help with this too.
>>> 
>>> Cheers,
>>> Geordie 
>>> Rick Mann <rmann at latencyzero.com <mailto:rmann at latencyzero.com>> schrieb am So. 7. Mai 2017 um 23:06:
>>> I'm trying to use apinotes for this third-party C library (call it "Lib.dylib"). It has an enum lgs_error_t:
>>> 
>>> typedef enum {
>>>     lgs_error_none = 0,
>>>     lgs_error_invalid_handle = -1,
>>>     lgs_error_null = -2,
>>>     lgs_error_invalid_parameter = -3,
>>>     lgs_error_invalid_operation = -4,
>>>     lgs_error_queue_full = -5
>>> } lgs_error_t;
>>> 
>>> So I wrote apinotes ("Lib.apinotes") that look like this, next to the .dylib, and part of my Xcode iOS app target:
>>> 
>>> Enumerators:
>>> # lgs_error_t
>>> 
>>> - Name: lgs_error_none
>>>   SwiftName: lgs_error_t.none
>>> - Name: lgs_error_invalid_handle
>>>   SwiftName: lgs_error_t.invalidHandle
>>> - Name: lgs_error_null
>>>   SwiftName: lgs_error_t.nullParameter
>>> - Name: lgs_error_invalid_parameter
>>>   SwiftName: lgs_error_t.invalideParameter
>>> - Name: lgs_error_invalid_operation
>>>   SwiftName: lgs_error_t.invalidOperation
>>> - Name: lgs_error_queue_full
>>>   SwiftName: lgs_error_t.queueFull
>>> 
>>> But this line of code fails:
>>> 
>>> var err: lgs_error_t = .nullParameter
>>> Type 'lgs_error_t' has no member 'nullParameter'
>>> 
>>> Am I missing something else?
>>> 
>>> > On May 4, 2017, at 16:55 , Douglas Gregor via swift-users <swift-users at swift.org <mailto:swift-users at swift.org>> wrote:
>>> >
>>> >
>>> >> On May 3, 2017, at 4:10 PM, Geordie J via swift-users <swift-users at swift.org <mailto:swift-users at swift.org>> wrote:
>>> >>
>>> >> Hi everyone,
>>> >>
>>> >> I’m about to start on another big project with Swift on Android and would like to annotate that JNI headers as much as possible before I do: specifically I’d like to make _Nonnull and CF_SWIFT_NAME annotations to the headers found in a user's jni.h.
>>> >>
>>> >> The question is: is it possible to annotate headers this without changing the original header files? Specifically I’m looking for an options that allows annotations in a separate file, probably one that is read when loading the package’s module.modulemap.
>>> >>
>>> >> I’d like to distribute the annotations in a SwiftPM package that also exposes the original (hopefully annotated) headers. Up until now I’ve been using Swift to override methods in code, but this isn’t as clean or extensible and I fear it may have other (particularly performance) implications.
>>> >>
>>> >> I guess the alternative would be to just maintain and distribute a modified version of jni.h with the annotations, but that would be a "last resort” option.
>>> >
>>> >
>>> > This is the role of API notes, which you can see here:
>>> >
>>> >       https://github.com/apple/swift/tree/master/apinotes <https://github.com/apple/swift/tree/master/apinotes>
>>> >
>>> > with some rough documentation-in-source here:
>>> >
>>> >       https://github.com/apple/swift-clang/blob/stable/lib/APINotes/APINotesYAMLCompiler.cpp <https://github.com/apple/swift-clang/blob/stable/lib/APINotes/APINotesYAMLCompiler.cpp>
>>> >
>>> >       - Doug
>>> >
>>> > _______________________________________________
>>> > swift-users mailing list
>>> > swift-users at swift.org <mailto:swift-users at swift.org>
>>> > https://lists.swift.org/mailman/listinfo/swift-users <https://lists.swift.org/mailman/listinfo/swift-users>
>>> 
>>> 
>>> --
>>> Rick Mann
>>> rmann at latencyzero.com <mailto:rmann at latencyzero.com>
>>> 
>>> 
>> 
>> _______________________________________________
>> swift-users mailing list
>> swift-users at swift.org <mailto:swift-users at swift.org>
>> https://lists.swift.org/mailman/listinfo/swift-users
> 
> _______________________________________________
> swift-users mailing list
> swift-users at swift.org
> https://lists.swift.org/mailman/listinfo/swift-users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.swift.org/pipermail/swift-users/attachments/20170512/ad745427/attachment.html>