Re: [wtp-dev] [vjet-dev] How do we get JSDT issues that have patches act

[John Peberdy <john@xxxxxxxxxx>]

On Tue, Jan 29, 2013 at 1:29 PM, Justin Early <earlyster@xxxxxxxxx> wrote:
[..]
> For this
>
> For our variant of JSDT [1] we disabled JSDT's problem checking and
> instead plan to use an optimized version of JSHint-eclipse [2].
>
>
> Do you see any issues with performance? JSDT has to parse the js file for
> search and code assist, then jshint has to parse and report issues.

There's a couple reasons we're leaning towards JSHint.

- The parsing code is complicated and obscure at least for me. There
seems to be multiple versions of the parser with different support wrt
semicolons. To improve the parsing code I would have to spend a lot of
time understanding it.
- For content assist and type inference we can be slightly more sloppy
than with problem checking - for type inference in the currently
edited file JSDT's parser is sufficient
- Content assist (for globals and global types) receives its data from
the NJSDoc algorithm so there's no reason to parse other files with
JSDT's parsers (except as fall back when NJSDoc doesn't know about the
type)
- I guess the main reason is that we differ ideologically about the
role of problem checking (VJET versus my variant of JSDT). My
impression is VJET believes it can have complete information about all
types (like Java for example). VJET shows warnings if a field or
member is unresolved. However JavaScript is very dynamic and my
experience building something similar for Scheme taught me this is
impossible in general and especially if one must constrain how
developers can write code to do that.


> Justin
>
>
> On Jan 29, 2013, at 10:16 AM, John Peberdy <john@xxxxxxxxxx> wrote:
>
> On Tue, Jan 29, 2013 at 11:15 AM, Justin Early <earlyster@xxxxxxxxx> wrote:
>
> Can you be more specific on this issue?
>
> the lack of handling semi-colons properly
>
>
> Both JSDT and VJET report spurious errors for this code
>
> someVar = 1 + 2
> ++someVar
>
> For our variant of JSDT [1] we disabled JSDT's problem checking and
> instead plan to use an optimized version of JSHint-eclipse [2].
>
>
> [1] https://bitbucket.org/nexj/webtools.jsdt.core/
> [2] http://github.eclipsesource.com/jshint-eclipse/
>
>
>
>
>
> Thanks,
>
> Justin
>
> Sent from my iPad
>
> On Jan 29, 2013, at 1:58 AM, Max Rydahl Andersen <max.andersen@xxxxxxxxxx>
> wrote:
>
> Hi guys,
>
> We got a bunch of fixes/patches submitted in bugzilla trying to get JSDT not
> fail on common available javascript libraries;
> especially in context of libraries popular in mobile tools.
>
> Similar errors still exist in VJET which has its own fork of JDST so i'm
> cc'ing vjet-dev on this topic too.
>
> Some of these issues and their patches are close to a year old...time to get
> these pushed in IMO.
>
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=366247 - NPE in
> CompilationUnitBinding#sourceMethod() (triggered by birt and other js
> libraries)
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=356446 - [exceptions] Errors
> running builder 'JavaScript Validator' on project with
> ArrayIndexOutOfBoundsException (Also fixes #344745)
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=344745 - [search]
> ArrayIndexOutOfBoundsException in DiskIndex during problem detection
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=350693 - [exceptions][parser]
> ArrayIndexOutOfBoundsException in Parser#consumeEmptyStatement on a
> javascript file when building project
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=357353 -
> ArrayIndexOutOfBoundsException in
> UnconditionalFlowInfo#markAsDefinitelyNonNull() with FieldReference
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=349020 - [validation] Eclipse
> complains about jquery minified JS files
>
> These makes it much more tolerable to use JSDT with javascript libraries.
>
> One issue we haven't found a way to solve yet is the lack of handling
> semi-colons properly - but would love to see it fixed but lack the info on
> how to update/modify the parser.
>
> But lets start with what we already got patches for - how do we get these
> issues with already existing patches fixed for Kepler ?
>
> /max
> _______________________________________________
> wtp-dev mailing list
> wtp-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/wtp-dev
>
> _______________________________________________
> wtp-dev mailing list
> wtp-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/wtp-dev
>
>
> --
> John Peberdy
> _______________________________________________
> wtp-dev mailing list
> wtp-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/wtp-dev
>
>
>
> _______________________________________________
> vjet-dev mailing list
> vjet-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/vjet-dev
>



--
John Peberdy