Closed drkstr101 closed 4 years ago
Here are the included updates so far
org.agileware.natural.common
has been repurposed.
The common
bundle was renamed to stepmatcher
. A new bundle named common
was
created in its place and given a major version bump (1.x -> 2.x). Both cucumber
and jbehave
depends on common
, while only cucumber
depends on stepmatcher
.There are no longer any planned feature updates for this branch, and so WIP flag has been removed. However, there is still a full round of testing that needs to be done to ensure the updates do not have any unintended consequences. It may be a good idea if we can distribute a release candidate to some users outside of Eclipse Marketplace, as there are many diverse setups and use cases I'm entirely unaware of.
src-gen/org/agileware/natural/cucumber/parser/antlr/internal/DebugInternalCucumber.g
.
This file may be loaded into ANTLRWorks for debugging. Just launch the executable jar
antlrworks-1.5.1.jar
located in the common
bundle. Note utility jars located in commons
are not on the classpath, however other bundles may put them on their own classpath by
referencing the bundle path in the build.properties. See `.tests' bundles for example.Updated for conflicts.
@rlogiacco Congrats on the new release!
All test cases in next
are now passing, and I believe the coverage is sufficient enough to begin manual testing of a release package today. I will do the first round of this myself so as not to waste anybody's time.
There are still a couple test failures in Jbehave, or actually one real issue that bubbles up to 5 test failures overall. This is caused by specific variations of the GivenScenarios
feature, which is in the spec, but I actually never found any usage of in my example stories. I am confident enough these issues in the ANTLR grammar are localized to the feature, and will not affect the overall structure of parsed AST.
Should we list / reference the issues that should get fixed with this PR so that they get automatically close upon merge?
Great idea!
I will start going through them and collecting them here.
Are we ready to merge this and start wrapping up a new release?
Are we aware each and every issue referencing this PR will get closed once we merge?
Nope! I was not aware!
This was meant to be a preliminary list for my reference as I build out more tests. I didn't realize these would all be closed automatically. I need to go through and pull some of these out.
edit
All referenced issues are those which may have been affected by changes in next, which turned out to be pretty much all of them. I wanted to go through each one and see exactly what might have changed the behavior, and how/why.
Are we ready to merge this and start wrapping up a new release?
I had accidentally included some unintended updates to the formatting2 when I merged from another branch which was causing test failures, but it should be good now. I just wanted to give it a quick run a few times to see if anything seems obviously broken.
Also, Cucumber
lang is now fully compliant ANTLR grammar (EG. no lexer/parser warnings during MWE2 workflow). This may have unintended affects in the UI however, so I will need to spend a little time smoke-screening it.
Seems ok at first glance: https://imgur.com/a/dsewf1I
But it seems formatting is still broken, even with the original formatter API. I may have done something to cause this however. I will investigate further.
We seem to be facing an up-hill-battle trying to get original CucumberJavaFormatter
working in latest Xtext, so rather than spinning cycles that will eventually be abandoned, they were spent on removing the last few blockers on the formatting2
api, which means it is now at least as capable as using the original formatter (in next), with a lot of room for custom fine tuning as needed/desired.
This implementation did however come with a slight trade off, in that the Xtext abstraction is now leaking into the AST. Or in other words, a slightly less ideal AST moel for the sole purpose of being easier to format.
An ideal implementation would likely come in the form of a core language model that had proper free-form multi-line text formatting support. This is unfortunately a project in and of itself.
Context: https://www.eclipse.org/forums/index.php?t=msg&th=1103866&goto=1828087&#msg_1828087
Next up, figure out WTH happened to stepmatching!
~I tested this awhile ago and was working fine, so I havent tested it since, thinking nothing I did would have changed anything~... but as I type this my brain just debugged the problem in the background... Step model was slightly changes. Stepmatcher must also be changed to follow suit. Should be fairly minor update.
@rlogiacco next
has been reset and beta
has been orphaned. I now have an alpha
branch that contains the previous next
, in case you would like me to open this PR back up. My intent however is to attempt to manually merge pieces from beta
-> next
, resubmitting the PR once I have a stable set of features in next
.
This branch is still under development. The WIP flag will be removed once it is ready for review.