Release notes for 1.4 RC
Scott Blum
I've put together some release notes for the 1.4 RC. They're complete except for some text from Joel about the breaking UIObject.styleName changes. I've included a build of the doc from the tip of the trunk because I included a bunch of references from the notes straight into the doc. Feedback much appreciated.
Scott
(FYI: the 1.4.99 reference is just a desk-build I did that I wanted to make sure wouldn't conflict with a real build.)
BobV
> Feedback much appreciated.
"Important changes in the behavior of existing GWT features."
- Change dot to colon, or remove the statement
"failed to warn at compile time about things that could cause problems
at runtime"
- failed to generate compile-time warnings about anticipated runtime errors.
"for code that works (or used to work) just fine"
- for code that used to work just fine.
"Bad code that worked before might not now"
- Anticipated user question: What if I want to return a collection
of types that don't derive from a common superclass?
"should have an associated gwt.typeArgs javadoc annotation"
- Add a hyperlink to the docs that talk about typeArgs
"warning can be suppressed via a module property."
- Is this documented somewhere? If not, specify the name of the property.
"cause UnsatisfiedLinkErrors that method is"
- Errors _when_ that method is
"importantly, the newstartup sequence"
- Missing a space in 'newstartup'
"removes an HTTP round-trip, startup latency about 33% faster."
- round-trip, decreasing startup latency by 33%.
"The above optimizations combined with ImageBundle, "
- These optimizations, when combined with ImageBundle (<-- make this
a link to an internal achor), make it possible for GWT-based
applications to load even faster. (Add a link to the mail sample on
the website?)
"easy to parse and format dates, time"
- For some reason 'format and parse' reads better to me, not sure why
"into a single "image strip","
- Provide a link to a generated image strip?
"requests into one: a single, permanently-cacheable image file"
- requests into a single, permanently-cacheable image request"
"as an Image widget or as pure HTML for inclusion in "
- Link to functions within Javadoc?
"help make the UI look better during startup, too."
- increase the robustness of your UI during startup
"To fix this problem"
- fixing this problem
"pre-initializing the width and height of each individual image ahead of time."
- "pre-initializing" implies "ahead of time".
"Image bundles do the same thing automatically. The dimensions of each
clipped image are computed automatically "
- Image bundles automatically compute the dimensions of each clipped
image during compilation...
"a fast, non-ugly user"
- s/non-ugly/elegant/
--
Bob Vawter
Google Web Toolkit Team
Scott Blum
Okay, well in some of the cases. Anything I don't mention below was a great find and I fixed. :) And updated release notes file is attached.
"Important changes in the behavior of existing GWT features."
- Change dot to colon, or remove the statement
It matches the pattern of the 1.1.0 notes to use a period there.
"failed to warn at compile time about things that could cause problems
at runtime"
- failed to generate compile-time warnings about anticipated runtime errors.
I think the first version has a more approachable tone. I dislike "anticipated" because it implies an actor; and the actor in this case can't be the old RPC system, because that was the problem, that it didn't anticipate runtime errors.
"for code that works (or used to work) just fine"
- for code that used to work just fine.
This is for parallelism with the previous parenthetical. That is, you might get a warning for code that "still works" but an error for code that "used to work" (but no longer works due to the error).
"Bad code that worked before might not now"
- Anticipated user question: What if I want to return a collection
of types that don't derive from a common superclass?
Answer #1: Stop wanting wrong things.
Answer #2: Use IsSerializable as the base type.
"should have an associated gwt.typeArgs javadoc annotation"
- Add a hyperlink to the docs that talk about typeArgs
Good call.
"warning can be suppressed via a module property."
- Is this documented somewhere? If not, specify the name of the property.
It's not documented, but I'm not sure we want to make it obvious or easy to turn the warning off.. we want people to just fix their code.
"into a single "image strip","
- Provide a link to a generated image strip?
It's a good idea, but it'd be too flaky since the relative URL to (say) Mail's image bundle resource would change unpredictably if we editted any of the images.
"requests into one: a single, permanently-cacheable image file"
- requests into a single, permanently-cacheable image request"
I kind of like Bruce's wording because it plays off 'dozens' against 'one'.
"as an Image widget or as pure HTML for inclusion in "
- Link to functions within Javadoc?
Nah, they'll find it if they go reading the ImageBundle doc.
"help make the UI look better during startup, too."
- increase the robustness of your UI during startup
I have to side with Bruce on this one. It's about looking good.
"To fix this problem"
- fixing this problem
"pre-initializing the width and height of each individual image ahead of time."
- "pre-initializing" implies "ahead of time".
I think Bruce may have been explicitly trying to make that sentence as laborious and redundant as possible to convey the idea as a visceral level. But aside from that, "pre-initializing" and "ahead of time" actually mean two different things here. The former refers to code the sets the width and height of images before they are requested from the server; the latter refers to the act of writing such code.
"Image bundles do the same thing automatically. The dimensions of each
clipped image are computed automatically "
- Image bundles automatically compute the dimensions of each clipped
image during compilation...
"a fast, non-ugly user"
- s/non-ugly/elegant/
Thanks!
Scott
John Tamplin
See the GWT issue tracker for the complete list of enhancements in this release.
Fixed Issues
See the GWT issue tracker for the complete list of bug fixes in this release.
--
John A. Tamplin
Software Engineer, Google
Scott Blum
Joel Webber
We've formalized the way that style names are handled in UIObject (which applies to all widgets as well). Style names have been separated into "base", "dependent", and "secondary" styles, which are detailed in the UIObject documentation. The interface is unchanged (get/setStyleName(), add/removeStyleName()), should not affect most widgets, and should bring some clarity to the way CSS is handled.
Scott Blum
We failed a number of unit tests after this change, which seems like a good indicator that user code might break. In fact, any time we have to change to a unit test to adapt to different behavior from the system, that should be a pretty good indicator that we might be breaking someone.
Joel Webber
Scott Blum
Bruce Johnson
Before we press go on the RC, I'd like to tweak the phrasing to be less "we do this" and more "GWT does this". Since GWT is such a collaboration among users, contributors, and Googlers, the current wording inadvertently implies that there's an "us" and a "them" when really there is only "us" :-) Using the product name removes that potential misunderstanding and is more accurate since it really is the product behavior that has changed and been enhanced.
Scott Blum
Would you mind doing the tweaking? We need to get final release notes to Bob ASAP so he can be ready to publish to the website. They can't go to the website as-is, because they contain links to our documentation as it exists in the distro. If we were also doing a doc push, it would be trivial to resolve this, but since we're not pushing doc, it seems like the links just need to be removed.
Bruce Johnson
Scott Blum
Bruce Johnson
Committed as r1167.