Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
An idea popped up on IRC today that I think we need to discuss so we can plan future releases. Namely, AJI (asciidoctor-java-integration) is expected to use Asciidoclet as is the maven plugin. Since Asciidoclet uses AJI we're going to run into a problem with releasing AJI and Asciidoclet as non-SNAPSHOTs. Either AJI will have to use an older version of Asciidoclet (one behind?) or we will have to figure out another strategy.
Thoughts on this? Is there a way we can release the two projects in tandem? Are there any other circular dependencies to worry about? Any other strategies that come to mind? John |
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
There certainly does seem to be a circular dependency here. I think the best way to do this is to use a version behind for doclet and aji. For maven and gradle I think it's less of an use because we'll use whatever is most current after aji releases.
On Tue, May 28, 2013 at 7:00 PM, johncarl81 [via Asciidoctor :: Discussion] <[hidden email]> wrote: An idea popped up on IRC today that I think we need to discuss so we can plan future releases. Namely, AJI (asciidoctor-java-integration) is expected to use Asciidoclet as is the maven plugin. Since Asciidoclet uses AJI we're going to run into a problem with releasing AJI and Asciidoclet as non-SNAPSHOTs. Either AJI will have to use an older version of Asciidoclet (one behind?) or we will have to figure out another strategy. |
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Hi, I would not consider this "problem" as circular dependencies, in a different environment but in my opinion almost the same Compilers Bootstraping
|
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
This post was updated on May 31, 2013; 3:27am.
Yes, I agree that this parallels compiler bootstrapping.
With that in mind, do you have any other solution ideas? I could see using some sort of shared project as the bootstrap, which would allow both projects to use a common version of Asciidoctor. However, I think this may be overkill for this situation and I still prefer the idea of AJI using one version behind of Asciidoclet. Thoughts? If we go with the one-version-behind idea, I can issue a release now of 0.1.2 so 0.1.3 of AJI will have a released version of Asciidoclet. |
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Well this weekend I am going to release the 0.1.3 version so, I think that you could release next week doclet with version 0.1.3 of AJI. Then for next release I will migrate all javadoc to AsciiDoc, and it will be rendered with version 0.1.3. I think that because of nature of javadoc it would not be necessary to use the latest version of AJI. What you think?
|
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Sounds perfect Alex. I'll release Asciidoclet after you release AJI.
John |
Loading... |
Reply to author |
Edit post |
Move post |
Delete this post |
Delete this post and replies |
Change post date |
Print post |
Permalink |
Raw mail |
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
Administrator
|
In reply to this post by johncarl81
I think that relying on a stable version of Asciidoclet is reasonable. This removes the circular dependency because we look at the Doclet as just another tool, not as an integrated part of the build/dependencies. -Dan On Wed, May 29, 2013 at 7:39 PM, johncarl81 [via Asciidoctor :: Discussion] <[hidden email]> wrote: Yes, I agree that this parallels compiler bootstrapping. ... [show rest of quote] Dan Allen | http://google.com/profiles/dan.j.allen
|
Free forum by Nabble | Edit this page |