Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 346960

Summary:	[validation] Unresolved variables validation is gone in 3.3m7
Product:	[WebTools] JSDT	Reporter:	Sergey <serge.che>
Component:	General	Assignee:	Project Inbox <jsdt.javascript-inbox>
Status:	NEW ---	QA Contact:	Chris Jaun <cmjaun>
Severity:	enhancement
Priority:	P3	CC:	cmjaun, lemartin, robinpelgrim
Version:	unspecified
Target Milestone:	Future
Hardware:	Macintosh
OS:	Mac OS X - Carbon (unsup.)
Whiteboard:

Description Sergey

2011-05-24 06:26:18 EDT

Build Identifier: 1.3.0.v201103031824-7F78FXPFBBoPbXRIcIgs0z0

When I updated to 3.3m7 of Eclipse WTP, one of the most valuable validation for me — Unresolved methods/variables/types — is gone. Is it possible to bring it back?

Reproducible: Always

Steps to Reproduce:
Try the following code:

"string".indexOf(someVar);

In Eclipse WTP 3.2 and 3.3m6 there’s an error "someVar cannot be resolved", which is gone in 3.3m7

Comment 1 Chris Jaun

2011-06-29 15:33:14 EDT

A decision was made for 3.2.4 and 3.3 to disable checking for unresolvable methods/variables/types.

This was done because the validation so often returned false positive results. It was one of the most complained about problems.

No work is planned in the near future to support this type of validation.

I have converted this bug to an enhancement to keep it on our radar for the future.

Comment 2 Robin

2011-06-30 08:42:14 EDT

A lot of true positive results are gone now too!!
And they where valuable to me.

For people complaining about the false positive results it was possible to
disable it (before the 'fix') anyway, so why remove these validations so rigorously?

Comment 3 Sergey

2011-07-03 07:21:24 EDT

I agree with @Robin: if it was possible to disable this validation option, why did you removed it at all? At least you can show a warning message that this feature is unstable and may produce many false positive results.

(In reply to comment #1)
> A decision was made for 3.2.4 and 3.3 to disable checking for unresolvable
> methods/variables/types.
> 
> This was done because the validation so often returned false positive results.
> It was one of the most complained about problems.
> 
> No work is planned in the near future to support this type of validation.
> 
> I have converted this bug to an enhancement to keep it on our radar for the
> future.

Comment 4 Martin

2011-08-01 13:21:16 EDT

(In reply to comment #1)
> A decision was made for 3.2.4 and 3.3 to disable checking for unresolvable
> methods/variables/types.
> 
> This was done because the validation so often returned false positive results.
> It was one of the most complained about problems.
> 
> No work is planned in the near future to support this type of validation.
> 
> I have converted this bug to an enhancement to keep it on our radar for the
> future.

Comment 5 Martin

2011-08-01 13:33:14 EDT

The unresolved variables was _the_ JSDT feature, making it possible to detect many bugs and typos without running the actual thing. There may've been some false positives, but still, if you know it's correct, you can just ignore it, if you don't know something is wrong, you can't fix it. And if you're really annoyed by these false positives, there was still the option to disable the unresolved variables check.

As I'm developing quite large projects, I take any static checking, even if it's not perfect. Please bring back the old semantic validator options, enabling us to write better JS. For now I went back to 3.2.3.