JSDT/Confcalls/Minutes 20160712

JSDT Weekly Syngc 12 July 2016

Attendees: Ilya, Patrik, , Victor, Angel, Gorkem

(JSDT Open Bugs: https://goo.gl/nPLxeF)

Status

Victor Fixed some selection issues, and triaged esprima issues, related to old JDK version. Snjezana provided some changes to be pushed Ilya: we recently get some feedback for JS in general. We're going to talk mainly about two issues: 1. Bug 356606 - [formatter] Unwanted alignment on Enter when inside array literal (since 2011), people are really pissed off. 2. Bug 419931 - Array initializer auto-formatting is a hideous, broken nightmare Formatter issue is one of the most important things that we should tackle, please, see also comments below the video: https://www.youtube.com/watch?v=UxGwu2adzIc There is a github patch , that tried to addess the problem, but is not working as it is expected (see: https://goo.gl/xNNXPN), as We should really work on these issues, and fix them for Neon.1, to improve the reputation of JSDT

Other issues

There is some issue building the sourceediting. It looks like the required orbit bundle is handled in wtp-parent/pom.xml line 49. So I'll have to find a correct source to download the Orbit bundle. (it looks like it works with http://download.eclipse.org/tools/orbit/downloads/drops/R20160520211859/repository/ URL for Orbit, however I'm not sure if it's correct from component versions point of view)

Updates

UPD. No, changing URL for Orbit didn't helped with the following errors: [ERROR] osgi.bundle,org.eclipse.draw2d,3.10.100.201605310205 [ERROR] osgi.bundle,org.eclipse.gef,3.11.0.201605310205

Debugger

Denis could help in switching to static parser, to handle the webkit protocol. Now we're using the dynamic parser, that is not backward compatible. The fix is about using the static parser, and is already tested to work in debugging angular js application Ilya pushed this doc: https://gist.github.com/ibuziuk/dac2d2701ab81be36ac98e8ad0f2d943 Now we have nodeJS runtime, and the idea is to add a Chromium Runtime hooked with webinspector propocol. Once we did this, we could even link our dynwebapp opened in that chrome instance. Please, see also Bug 359436 - Investigate defining browsers / runtimes to use for debugging That give us the possibility to add new runtimes. For chromium we're using it to open html pages or other in debug. the problem: Debugger works now, but it is unhandy, as you have to specify manually all data, like debugger ports and other parameters. (Q) What about other browsers, other than Chromium? - For Neon.1 we should focus on Chromium - Opera is Chromium , so there is no pb - proxy for IE, and Firefox should be added later (Q) Should we check for the existence of Chromium ? - Y, it seems a copy og the NodeJS one, and it seems there are no problem with that. (inspired from JDT runtime configuration libraries) -- Note: we don't have a server adapter for NodeJS (and probably is not possible to have one)

   what we can do, is to have a server adapter for [express server adapter]

(Q) What if I have an already running browser ? - in NodeJS, I can access it! - if you've Chromium already running, and you try to hook it, it seems it cannot lock a debug port (looks like a Chromium bug) - So, seems that: no chromium instances running -> I can set the debugger port - maybe it is that way because of Chrome having one process per tab. - there is a Netbeans plugin that seems able to access Chromium in that way, but it is an old plug-in

Angel was looking at JSDT search and indexing stuffs, a lot of readings The array auto-complete looks like a pretty easy fix. Good idea to fix it

Oomph

Patrik: working on Oomph installer for JSDT, will complete this week. Then I will link introductory docs to get started to contribute to JSDT, to understand: how sourceediting works, where to put hands to fix something. Ilya: tracking this, maybe via email to wtp-dev Patrik: up to send a mailing list about Oomph.