Re: [lang] Thread safety annotations

Oh that's a pretty decent advantage there actually. On 29 November 2016 at 22:07, Gary Gregory wrote: > On Tue, Nov 29, 2016 at 8:04 PM, Matt Sicker wrote: > > > That's pretty neat. So it looks like two competing styles of annotations > > here would be

Re: [lang] Thread safety annotations

On Tue, Nov 29, 2016 at 8:04 PM, Matt Sicker wrote: > That's pretty neat. So it looks like two competing styles of annotations > here would be separate annotations for each concern, a single annotation > with an enum, or even separate annotations with boolean values or enums >

Re: [lang] Thread safety annotations

That's pretty neat. So it looks like two competing styles of annotations here would be separate annotations for each concern, a single annotation with an enum, or even separate annotations with boolean values or enums (though that gets a little more complex come to think of it). If only the JDK

Re: [lang] Thread safety annotations

And an example: /** * HTTP/2 compatible extension of {@link RequestConnControl}. * * @since 5.0 */ @Contract(threading = ThreadingBehavior.IMMUTABLE) public class H2RequestConnControl extends RequestConnControl { ... Gary On Tue, Nov 29, 2016 at 6:40 PM, Gary Gregory

Re: [lang] Thread safety annotations

FYI, this is how we're going to do it in HttpComponents 5: https://svn.apache.org/repos/asf/httpcomponents/httpcore/trunk/httpcore5/src/main/java/org/apache/hc/core5/annotation/ThreadingBehavior.java Gary On Tue, Nov 29, 2016 at 2:20 PM, Gary Gregory wrote: > On Tue,

Re: [lang] Thread safety annotations

On Tue, Nov 29, 2016 at 1:59 PM, Benedikt Ritter wrote: > I agree with Oliver. I don't think we have consensus about if and how we > want to implement this. Does it even make sense to try to document this? > What value does such a documentation have, when it is wrong 50% of

Re: [lang] Thread safety annotations

I agree with Oliver. I don't think we have consensus about if and how we want to implement this. Does it even make sense to try to document this? What value does such a documentation have, when it is wrong 50% of the time? I just sumed up the feedback I got from one (!) user at the conference. I

Re: [lang] Thread safety annotations

On Tue, Nov 29, 2016 at 1:12 PM, Oliver Heger wrote: > > > Am 28.11.2016 um 22:28 schrieb Gary Gregory: > > WRT licensing, I re-implemented the annotations in git master. Notice > that > > all Javadocs are different and much simpler for now. Feel free to write > >

Re: [lang] Thread safety annotations

Am 28.11.2016 um 22:28 schrieb Gary Gregory: > WRT licensing, I re-implemented the annotations in git master. Notice that > all Javadocs are different and much simpler for now. Feel free to write > more. As the length of this thread shows, using annotations for the purpose of documenting

Re: [lang] Thread safety annotations

WRT licensing, I re-implemented the annotations in git master. Notice that all Javadocs are different and much simpler for now. Feel free to write more. Gary On Mon, Nov 28, 2016 at 12:55 PM, Gary Gregory wrote: > On Mon, Nov 28, 2016 at 12:18 PM, Matt Benson

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 12:18 PM, Matt Benson wrote: > I hoped to stay out of this. :) Points: > > * "provided" scope is another Maven mechanism that compiles against a > given dependency but does not pull the dependency in as a runtime > dependency. I personally prefer it

Re: [lang] Thread safety annotations

I hoped to stay out of this. :) Points: * "provided" scope is another Maven mechanism that compiles against a given dependency but does not pull the dependency in as a runtime dependency. I personally prefer it to marking a dependency optional. * For annotations with only classfile retention, a

Re: [lang] Thread safety annotations

Groovy had to change the license of its documentation from CC-A 3.0 to the Apache License during incubation: https://issues.apache.org/jira/browse/LEGAL-167 http://markmail.org/message/2e7tehlwtpx625q4 https://issues.apache.org/jira/browse/GROOVY-7470 So I guess Commons is probably not allowed

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 7:15 AM, Jochen Wiedmann wrote: > On Mon, Nov 28, 2016 at 4:06 PM, sebb wrote: > > > The code would not run without the JCIP jar. > > Are there licensing issues regarding that jar? > Hm, according to

Re: [lang] Thread safety annotations

On 28 November 2016 at 15:15, Jochen Wiedmann wrote: > On Mon, Nov 28, 2016 at 4:06 PM, sebb wrote: > >> The code would not run without the JCIP jar. > > Are there licensing issues regarding that jar? I think there were, but regardless, the jar

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 4:13 PM, sebb wrote: > Either you have to allow a separate release cycle for that jar. > Or you have to release a new version of the jar with every LANG release. > The content will likely be identical for each one unless the Javadoc is > tweaked. I'd

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 4:06 PM, sebb wrote: > The code would not run without the JCIP jar. Are there licensing issues regarding that jar? Jochen -- The next time you hear: "Don't reinvent the wheel!"

Re: [lang] Thread safety annotations

On 28 November 2016 at 13:12, Jochen Wiedmann wrote: > On Mon, Nov 28, 2016 at 12:08 PM, sebb wrote: > >> No, I'm suggesting that the annotations become an independent Commons >> component. >> If we don't want these used outside Commons (probably a

Re: [lang] Thread safety annotations

On 28 November 2016 at 13:10, Jochen Wiedmann wrote: > On Mon, Nov 28, 2016 at 12:03 PM, sebb wrote: >> Yes, we started with the ones from JCIP, but these had runtime >> retention, which caused issues. > > What kind of problems, if I may ask? The

Re: [lang] Thread safety annotations

On Mon, 28 Nov 2016 14:12:48 +0100, Jochen Wiedmann wrote: On Mon, Nov 28, 2016 at 12:08 PM, sebb wrote: No, I'm suggesting that the annotations become an independent Commons component. If we don't want these used outside Commons (probably a good idea) then we can make this

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 12:08 PM, sebb wrote: > No, I'm suggesting that the annotations become an independent Commons > component. > If we don't want these used outside Commons (probably a good idea) > then we can make this clear through Javadoc and package naming etc. > > The

Re: [lang] Thread safety annotations

On Mon, Nov 28, 2016 at 12:03 PM, sebb wrote: > Yes, we started with the ones from JCIP, but these had runtime > retention, which caused issues. What kind of problems, if I may ask? Thanks, Jochen - To

Re: [lang] Thread safety annotations

On 28 November 2016 at 01:55, Gary Gregory wrote: > Are you suggesting [lang] becomes a multi-module project? No, I'm suggesting that the annotations become an independent Commons component. If we don't want these used outside Commons (probably a good idea) then we can

Re: [lang] Thread safety annotations

Yes, we started with the ones from JCIP, but these had runtime retention, which caused issues. I e-mailed JCIP asking if the retention could be changed, but no joy. We therefore created our own from elsewhere. Not sure why we eventually dropped them. On 28 November 2016 at 04:06, Matt Sicker

Re: [lang] Thread safety annotations

Ah, ok, what I have in there now is in our own package. Gary On Nov 27, 2016 8:06 PM, "Matt Sicker" wrote: > Well, HttpClient used to have the same (or similar) thread safety > annotations for documentation purposes, and I think they were just custom > annotations and not the

Re: [lang] Thread safety annotations

Well, HttpClient used to have the same (or similar) thread safety annotations for documentation purposes, and I think they were just custom annotations and not the javax ones. On 27 November 2016 at 20:50, Gary Gregory wrote: > On Nov 27, 2016 6:46 PM, "Matt Sicker"

Re: [lang] Thread safety annotations

On Nov 27, 2016 6:46 PM, "Matt Sicker" wrote: > > I'd be in favor of just including this as an internal annotation similar to > how it's done in HttpClient. What does that mean specifically? FWIW, I think those annotations might be gone in the current svn trunk. Gary > > On 27

Re: [lang] Thread safety annotations

Oh, and JSR 305 does not actually specify any concurrency annotations. Gary On Nov 27, 2016 6:46 PM, "Matt Sicker" wrote: > I'd be in favor of just including this as an internal annotation similar to > how it's done in HttpClient. > > On 27 November 2016 at 20:44, Gary

Re: [lang] Thread safety annotations

Also note that the JCIP code uses its own package AND declares the annotations with the RUNTIME retention policy. Gary On Nov 27, 2016 6:39 PM, "Matt Sicker" wrote: > If you're going to share the annotations between projects, why not use a > JSR package for it like the google

Re: [lang] Thread safety annotations

I'd be in favor of just including this as an internal annotation similar to how it's done in HttpClient. On 27 November 2016 at 20:44, Gary Gregory wrote: > I do not think we should use an official Java package because the JSR 305 > page says "The Executive Committee

Re: [lang] Thread safety annotations

I do not think we should use an official Java package because the JSR 305 page says "The Executive Committee voted to list this JSR as dormant in May 2012." Gary On Nov 27, 2016 6:39 PM, "Matt Sicker" wrote: > If you're going to share the annotations between projects, why not

Re: [lang] Thread safety annotations

If you're going to share the annotations between projects, why not use a JSR package for it like the google one? I thought the idea of embedding the annotations in [lang] was to prevent the need for the dependency and to use it internally, not as a public annotation for other projects to use. On

Re: [lang] Thread safety annotations

Are you suggesting [lang] becomes a multi-module project? Gary On Nov 27, 2016 5:01 PM, "sebb" wrote: > On 28 November 2016 at 00:11, Gary Gregory wrote: > > On Sun, Nov 27, 2016 at 4:09 PM, sebb wrote: > > > >> On 27 November 2016

Re: [lang] Thread safety annotations

On 28 November 2016 at 00:11, Gary Gregory wrote: > On Sun, Nov 27, 2016 at 4:09 PM, sebb wrote: > >> On 27 November 2016 at 23:51, Gary Gregory wrote: >> > On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: >> > >>

Re: [lang] Thread safety annotations

On 28 November 2016 at 10:51, Gary Gregory wrote: > On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > >> On 27 November 2016 at 22:31, Gary Gregory wrote: >> > On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote: >> > >>

Re: [lang] Thread safety annotations

On Sun, Nov 27, 2016 at 4:09 PM, sebb wrote: > On 27 November 2016 at 23:51, Gary Gregory wrote: > > On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > > > >> On 27 November 2016 at 22:31, Gary Gregory > wrote: >

Re: [lang] Thread safety annotations

On 27 November 2016 at 23:51, Gary Gregory wrote: > On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > >> On 27 November 2016 at 22:31, Gary Gregory wrote: >> > On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote: >> > >>

Re: [lang] Thread safety annotations

On Sun, Nov 27, 2016 at 3:51 PM, Gary Gregory wrote: > On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > >> On 27 November 2016 at 22:31, Gary Gregory >> wrote: >> > On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote:

Re: [lang] Thread safety annotations

On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > On 27 November 2016 at 22:31, Gary Gregory wrote: > > On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote: > > > >> On 27 November 2016 at 22:03, Gary Gregory > wrote: >

Re: [lang] Thread safety annotations

On 28 November 2016 at 10:41, sebb wrote: > On 27 November 2016 at 22:31, Gary Gregory wrote: >> It is _because_ they are NOT built-in the language or JRE that we are >> proposing they belong in [lang]. >> >> Since we are providing the annotation with

Re: [lang] Thread safety annotations

Please see https://issues.apache.org/jira/browse/LANG-1291 and git master main and unit tests. It's pretty neat that you can only have a compile-time dependency and not a runtime dependency as the unit tests show. Gary On Sun, Nov 27, 2016 at 3:41 PM, sebb wrote: > On 27

Re: [lang] Thread safety annotations

On 27 November 2016 at 22:31, Gary Gregory wrote: > On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote: > >> On 27 November 2016 at 22:03, Gary Gregory wrote: >> > These annotations are the SAME as have been published all over the

Re: [lang] Thread safety annotations

On Sun, Nov 27, 2016 at 2:14 PM, sebb wrote: > On 27 November 2016 at 22:03, Gary Gregory wrote: > > These annotations are the SAME as have been published all over the place, > > so I do not think we need a PR for review. Reviewing the code in the repo

Re: [lang] Thread safety annotations

On 27 November 2016 at 22:03, Gary Gregory wrote: > These annotations are the SAME as have been published all over the place, > so I do not think we need a PR for review. Reviewing the code in the repo > should be enough, but I'm not sure _what_ there is to review really.

Re: [lang] Thread safety annotations

These annotations are the SAME as have been published all over the place, so I do not think we need a PR for review. Reviewing the code in the repo should be enough, but I'm not sure _what_ there is to review really. I've already posted a link to the sources (a pastebin link). I'll create a Jira

Re: [lang] Thread safety annotations

Hi, Gary Gregory schrieb am So., 27. Nov. 2016 um 21:01 Uhr: > So let's start with the CLASS retention annotations first. I'll commit > later today. > Maybe create a GitHub PR so people can have a look before it goes to master? Benedikt > > Gary > > On Nov 27, 2016

Re: [lang] Thread safety annotations

So let's start with the CLASS retention annotations first. I'll commit later today. Gary On Nov 27, 2016 9:50 AM, "Matt Sicker" wrote: > I feel like documenting thread safety in such a low level, widely used > library like [lang] is a great idea. Whether it uses annotations,

Re: [lang] Thread safety annotations

My view is that it might be challenging to do and keep the information accurate over time, but that this is a worthy goal nonetheless. Having the annotations around will allow for anyone to do this in a standard way. Coming up with a Javadoc only way to do this is too much of a pain IMO. Gary On

Re: [lang] Thread safety annotations

I feel like documenting thread safety in such a low level, widely used library like [lang] is a great idea. Whether it uses annotations, javadocs, or some sort of informal javadoc comments isn't too big a deal to me, but providing some guarantees of thread safety or lack thereof would be

Re: [lang] Thread safety annotations

I thought we decided to abstain from documenting thread safety since it would get outdated quickly anyway. Gary Gregory schrieb am So., 27. Nov. 2016 um 09:11 Uhr: > But if we then want a runtime version we had an odd packaging with CLASS > retention in .concurrent. and

Re: [lang] Thread safety annotations

But if we then want a runtime version we had an odd packaging with CLASS retention in .concurrent. and RUNTIME in .concurrent.runtime. I might be in YAGNI territory here... Gary On Sat, Nov 26, 2016 at 10:07 PM, Matt Sicker wrote: > I think adding the additional

Re: [lang] Thread safety annotations

I think adding the additional clazz/runtime package would be a little overkill if you only include one packaging option for now. On 27 November 2016 at 00:04, Gary Gregory wrote: > On Nov 26, 2016 8:40 PM, "Matt Sicker" wrote: > > > > I can think of a

Re: [lang] Thread safety annotations

On Nov 26, 2016 8:40 PM, "Matt Sicker" wrote: > > I can think of a couple use cases for runtime retention for thread safety > annotations, but they're not specific to commons lang. > > 1. A framework could choose what type of data structure or algorithm to use > at runtime based

Re: [lang] Thread safety annotations

I can think of a couple use cases for runtime retention for thread safety annotations, but they're not specific to commons lang. 1. A framework could choose what type of data structure or algorithm to use at runtime based on the presence of which thread safety annotation is on. Same for

Re: [lang] Thread safety annotations

On Sat, Nov 26, 2016 at 3:54 PM, sebb wrote: > On 26 November 2016 at 19:25, Gary Gregory wrote: > > On a [crypto] thread, I mentioned http://pastebin.com/RKPGGdJ9 to add > the > > "classic" four javax.annotation.concurrent annotations to two packages >

Re: [lang] Thread safety annotations

On 26 November 2016 at 19:25, Gary Gregory wrote: > On a [crypto] thread, I mentioned http://pastebin.com/RKPGGdJ9 to add the > "classic" four javax.annotation.concurrent annotations to two packages > (.clazz and .runtime) in Commons Lang for CLASS and RUNTIME retentions.

[lang] Thread safety annotations

On a [crypto] thread, I mentioned http://pastebin.com/RKPGGdJ9 to add the "classic" four javax.annotation.concurrent annotations to two packages (.clazz and .runtime) in Commons Lang for CLASS and RUNTIME retentions. You'd use the CLASS version to avoid a hard dependency on [lang]. RUNTIME to