'. '


From APIDesign

Jump to: navigation, search

I'm sure you've already tried, but if you could try just a little harder to remember the API change or problem that you express in the "Unconsciousness" Note, I think this will be a stronger point.

--Dmkoelle 02:18, 23 April 2008 (UTC)

You talk about the difference between fast track and normal reviews, but not who gets to choose which is used and why. That might be helpful for people thinking about implementing such a thing at their office.

  • Read more at [1] API Reviews Page], it is more appropriate on the web than in the book, I think.

It was very interesting to learn your secret about NetBeans' API review process being exactly the same as Sun's.

The "Working Backwards" concept is really interesting and I am glad you mention it. As an aside, I came across it in the NetBeans Wiki by accident, because I was reading a press release about all the great features of the new version of NetBeans. I thought, wow, I didn't know 6.0.1 could do all of this -- but it was a "working backwards" press release about the unreleased 6.5 or 7.0 version!

I very strongly agree that without Geertjan there would be few people using the platform today, as I am one such person!

--TomWheeler Wed Apr 23 20:38:48 CDT 2008

The section "NetBeans Javadoc Extensions" seems a bit too NetBeans-specific.

--AdamDingle 17:14, 20 April 2008 (UTC)

Calling this section "Big Brother" sounds quite ominous.

Regarding the code coverage tool: does it really help with the non-determinism of multiple threads? I wasn't aware that such a thing could be tested thoroughly with confidence.

"Note: NetBeans Javadoc", paragraph 2: "The Javadoc is built..." -- to which Javadoc are you referring?

--Dmkoelle 02:23, 23 April 2008 (UTC)

Perhaps better to talk about <api> and <usecase> as about elements, not tags. The XML spec makes a clear difference between tags and elements.

"feel free to point your browser to": missing link: EnhancedJavadoc, SignatureTests

--AndreiBadea 09:40, 23 April 2008 (UTC)

"This kind of coverage is unmeasurable" is probably not strictly correct. It probably is measurable somehow, but simply not practical to do so.

It would be interesting to know what other "platform type" projects (such as Eclipse RCP) do to document and test compatibility between versions.

--TomWheeler Wed Apr 23 20:38:48 CDT 2008

Done: 4ac1d0cf7071

"However, if you ask for reasonable coverage of code lines changed by the patch, you may be confident that at least the basic functionality is acceptable." True, but it's quite possible that it broke something else in another part of the system. Having good test coverage across the entire application is the best way to detect/prevent that.

--TomWheeler Wed Apr 23 20:38:48 CDT 2008

Personal tools