COM support: Added significant enhancements. #397

dhakehurst · 2015-02-09T13:29:22Z

Added significant enhancements to the COM support.
Most of this is under the com.sun.jna.platform.win32.COM.util package, with a few additions to the com.sun.jna.platform.win32.COM package.

there are tests added (which I think cover all the added enhancements), plus we have now been using it extensively for 3 months.

Added proxy and annotations to assist easy use of COM. Added java friendly version of some COM interfaces. Added Tests.

COM threads for different COM object access. Explicit termination of COM thread necessary, finalizer use is not recommended as reliable. Corrected signature (IID -> REFIID) in IDispatch.

COM Event Callback stuff

… in dispatch.c, the parameter is used under WIN32 build.

Added source artifacts

possible. Required a change to Variant, to enable size of array of VariantArg to be set. Might be a cleaner way to do this, but it could break the existing API of VariantArg.

updates to handling callback arguments

dblock · 2015-02-09T16:28:07Z

Looks great, would you please update https://github.com/twall/jna/blob/master/CHANGES.md (i think it needs a "Next" section).

dblock · 2015-02-09T16:29:49Z

contrib/platform/src/com/sun/jna/platform/win32/Ole32.java

@@ -267,4 +267,62 @@ HRESULT CoCreateInstance(GUID rclsid, Pointer pUnkOuter, int dwClsContext,
 	 */
    void CoTaskMemFree(Pointer pv);

+    /**
+     * Retrieves a pointer to the default OLE task memory allocator


Missing a period :)

added a period :-)

dblock · 2015-02-09T16:36:05Z

This is some nice solid work here, @dhakehurst, hats off. I made a couple of cosmetic comments, feel free to do nothing about them. I will be happy to merge after the CHANGES update.

dhakehurst · 2015-02-10T09:16:21Z

Hi,
Not quite sure what to modify in the CHANGES.md file!

dblock · 2015-02-10T12:05:44Z

I think CHANGES needs something like what you wrote above:

Use of interfaces and annotations to provide easier implementation of COM interfaces (uses InvocationHandler).
Support for COM event callbacks (this was particularly tricky, very happy I got it to work).
Support for COM interface discovery by iteration over the RunningObjectTable.

Whatever else you want to write, maybe sprinkle it with specific class names since not everybody uses Windows (eg. Added com.sun.jna......).

I think you might want to squash these commits, there's a lot of small stuff going on, it will be hard to track history. However I don't feel strongly about this, you decide.

I don't think the eclipse or maven project should be committed. For one I personally prefer to track it outside of JNA with my own settings. Please remove it. Unless @twall feels otherwise. Or we can discuss adding those in a separate PR?

dblock · 2015-02-13T12:01:56Z

@dhakehurst LMK if you need help.

added changes to CHANGES.md (fixed a bug in ProxyObject)

dhakehurst · 2015-02-20T20:18:27Z

Have added some text to the CHANGES file.
I need the maven build in my version of the repo, no idea how to mark it not to be included in the pull request!

dblock · 2015-02-20T23:10:17Z

Can you tell us why you need the maven stuff? I understand you might prefer Maven, but this project doesn't use it to build.

To exclude it:

Move the maven folder(s) out of the way (like cut and paste elsewhere).
Add maven to .gitignore, git add --all and git commit (as if the maven folder was git rm-ed).
Put the maven folder back, it won't be tracked since it's in .gitignore.

dblock · 2015-02-20T23:10:36Z

And thanks :) This is a LOT of stuff, I am willing to help in any way I can to get this in.

dblock · 2015-02-20T23:11:02Z

CHANGES.md

@@ -7,6 +7,10 @@ Release 4.2

 Features
 --------
+* Significant enhancements to COM support under com.sun.jna.platform.win32.COM.util, - [@dhakehurst](https://github.com/dhakehurst)


You have an extra , at the end of this line before your name.

dhakehurst · 2015-03-04T09:38:19Z

The maven stuff is used by our Jenkins server to do the build. It need to be maven so that it integrates with the rest of the jenkins builds (up/down stream stuff).

Are you sure that no one else would be interested in maybe building with maven? Is it very wrong to have an alternative build approach in the repository? (it doesn't yet build the native stuff, but it could do with a bit more work.)

dblock · 2015-03-04T11:22:57Z

@twall Can comment about maven, but I am 99% sure, we have hard time maintaining one build mess ;) If you want to suggest building with maven I would suggest splitting it into a separate PR.

twall · 2015-03-04T11:49:24Z

While I’m not against a more mavenized build, the one we’ve got is a bit of a pain (most likely due to the three of us who usually do releases are not well versed in that space, or at least the providers we’re using).

We just need to be clear about what we’re getting out of it and that we’re able to maintain it.

I do all builds with ant (including all the native builds), then when time comes to put out a release either db, myself, or drollo will follow the maven release steps to get it out to the release repo. I’d love to have a more push-button approach to releases, though.

I agree that any maven build proposal should be in its own PR.

Thanks again for this COM work!

On Mar 4, 2015, at 6:23 AM, Daniel Doubrovkine (dB.) @dblockdotorg notifications@github.com wrote:

@twall Can comment about maven, but I am 99% sure, we have hard time maintaining one build mess ;) If you want to suggest building with maven I would suggest splitting it into a separate PR.

—
Reply to this email directly or view it on GitHub.

dhakehurst · 2015-03-04T12:29:07Z

I am not suggesting you move jna to maven.

However, I need the maven stuff because we build with maven. Is there a way for me to keep the maven folder in our clone of the repo, but not export it to you in the PR? There is none I know of.

Can I suggest that you simply exclude the maven folder when you do the merge. It is totally self contained, nothing outside it references into it.

twall · 2015-03-04T13:31:11Z

That’s acceptable. We can either do the merge manually or remove the maven folder after the merge.

On Mar 4, 2015, at 7:29 AM, Dr. David H. Akehurst notifications@github.com wrote:

I am not suggesting you move jna to maven.

However, I need the maven stuff because we build with maven. Is there a way for me to keep the maven folder in our clone of the repo, but not export it to you in the PR? There is none I know of.

Can I suggest that you simply exclude the maven folder when you do the merge. It is totally self contained, nothing outside it references into it.

—
Reply to this email directly or view it on GitHub.

dblock · 2015-03-04T14:22:30Z

I'll take care of the merge, but I will need some time to get to this.

bhamail · 2015-03-04T14:53:35Z

There are some known problems with the existing ant build (along the lines of dependent targets not working as expected “depending” on how/when invoked). This has lead to incorrect binaries being released in the past. I mention this because I’m open to any solution to that sort of problem (and a maven build might be one). I’ve thought about doing a fully mavenized build before, but I couldn’t see a clean way to further modularize the various native builds w/out a bunch of “moves” and/or kludges. I’m interested to hear more about how you’re approaching this.

Dan

On Mar 4, 2015, at 6:49 AM, Timothy Wall notifications@github.com wrote:

While I’m not against a more mavenized build, the one we’ve got is a bit of a pain (most likely due to the three of us who usually do releases are not well versed in that space, or at least the providers we’re using).

We just need to be clear about what we’re getting out of it and that we’re able to maintain it.

I do all builds with ant (including all the native builds), then when time comes to put out a release either db, myself, or drollo will follow the maven release steps to get it out to the release repo. I’d love to have a more push-button approach to releases, though.

I agree that any maven build proposal should be in its own PR.

Thanks again for this COM work!

On Mar 4, 2015, at 6:23 AM, Daniel Doubrovkine (dB.) @dblockdotorg notifications@github.com wrote:

@twall Can comment about maven, but I am 99% sure, we have hard time maintaining one build mess ;) If you want to suggest building with maven I would suggest splitting it into a separate PR.

—
Reply to this email directly or view it on GitHub.

—
Reply to this email directly or view it on GitHub #397 (comment).

dblock · 2015-03-16T14:29:05Z

I have merged the code from here minus any maven stuff via d6675fa cause I want to move forward. I got a bunch of failures in the tests, which I will look at and open as bugs - it's very likely that these are things missing from my environment.

Please make small PRs, looking forward to more.

dblock · 2015-03-16T15:09:46Z

Opened #406, #407 and #408.

…cess#397) Motivation: We did not correctly escape chars when shading is used. Modifications: - Update jni-utils to fix bug Result: Fixes escaping Related: - netty/netty-jni-util#13 - netty/netty#12358

dhakehurst added 30 commits October 21, 2014 17:33

COM: Added Running Object Table bits.

f279e84

Added proxy and annotations to assist easy use of COM. Added java friendly version of some COM interfaces. Added Tests.

Added pom files for building eclipse p2-repository, features and bundles

c5c8da5

Added eclipse project files

c153e96

renaming packages and interfaces in new COM bits

fe0ed1c

Added more eclipse files

0f399d5

Added .gitignore

06c884a

Calls to COM objects via ...utils should now all run on a single thread.

cdff360

Improve COM.utils tests

39e14d6

rename eclipse folder to maven folder

58054dd

change MsWord example to use new com stuff

209bd2f

Adding support for COM event callbacks

2ccf522

Adding support for COM Event callbacks

1141286

COM Event Callback support

00e6de5

COM.utils.Factoy should not be an static INSTANCE, may want different

046169f

COM threads for different COM object access. Explicit termination of COM thread necessary, finalizer use is not recommended as reliable. Corrected signature (IID -> REFIID) in IDispatch.

COM/COM.utils Copyright header modification

f368ad4

COM Event Callback stuff

Now supporting Event callbacks

f5c4bd5

Fix: Callback events executed on their own thread

02d7d3d

COM.utilÖ updates

048ed54

Remove UNUSED_ENV macro from Java_com_sun_jna_Native_getWindowHandle0…

c4e8aab

… in dispatch.c, the parameter is used under WIN32 build.

enable comparison of COM.util objects

b42ba9d

Iterator required 'remove' to be implemented for < Java 1.8

b53fd3a

Added source artifacts

comment out inclusion of testlib

b5c542b

comment out dependency on missing testlib

81ffec8

COM.util: bug fix: arguments to callback events now converted where

4cd8d30

possible. Required a change to Variant, to enable size of array of VariantArg to be set. Might be a cleaner way to do this, but it could break the existing API of VariantArg.

...COM.util: proxy must AddRef on COM handle and Release it

22915e0

updates to handling callback arguments

..COM.utils: maven build pom update

581cc00

added parameter to pom to allow commandline additional.pluginRepository

669697a

it needs a vaild default

58ea371

pom update

88c9a5f

maven build: update

0cdd1cb

dblock reviewed Feb 9, 2015
View reviewed changes

minor modifications

ce7a463

dhakehurst added 2 commits February 20, 2015 09:01

...COM.util:

293a62e

added changes to CHANGES.md (fixed a bug in ProxyObject)

Merge branch 'master' of https://github.com/twall/jna

81ba1be

dblock reviewed Feb 20, 2015
View reviewed changes

updated full stops

57aeb2f

dblock closed this Mar 16, 2015

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

COM support: Added significant enhancements. #397

COM support: Added significant enhancements. #397

dhakehurst commented Feb 9, 2015

dblock commented Feb 9, 2015

dblock Feb 9, 2015

dhakehurst Feb 10, 2015

dblock commented Feb 9, 2015

dhakehurst commented Feb 10, 2015

dblock commented Feb 10, 2015

dblock commented Feb 13, 2015

dhakehurst commented Feb 20, 2015

dblock commented Feb 20, 2015

dblock commented Feb 20, 2015

dblock Feb 20, 2015

dhakehurst commented Mar 4, 2015

dblock commented Mar 4, 2015

twall commented Mar 4, 2015

dhakehurst commented Mar 4, 2015

twall commented Mar 4, 2015

dblock commented Mar 4, 2015

bhamail commented Mar 4, 2015

dblock commented Mar 16, 2015

dblock commented Mar 16, 2015

COM support: Added significant enhancements. #397

COM support: Added significant enhancements. #397

Conversation

dhakehurst commented Feb 9, 2015

dblock commented Feb 9, 2015

dblock Feb 9, 2015

Choose a reason for hiding this comment

dhakehurst Feb 10, 2015

Choose a reason for hiding this comment

dblock commented Feb 9, 2015

dhakehurst commented Feb 10, 2015

dblock commented Feb 10, 2015

dblock commented Feb 13, 2015

dhakehurst commented Feb 20, 2015

dblock commented Feb 20, 2015

dblock commented Feb 20, 2015

dblock Feb 20, 2015

Choose a reason for hiding this comment

dhakehurst commented Mar 4, 2015

dblock commented Mar 4, 2015

twall commented Mar 4, 2015

dhakehurst commented Mar 4, 2015

twall commented Mar 4, 2015

dblock commented Mar 4, 2015

bhamail commented Mar 4, 2015

dblock commented Mar 16, 2015

dblock commented Mar 16, 2015