[jira] [Created] (THRIFT-4806) Convert NodeJS Library to typescript

2019-02-16 Thread Brian Forbis (JIRA)
Brian Forbis created THRIFT-4806:


 Summary: Convert NodeJS Library to typescript
 Key: THRIFT-4806
 URL: https://issues.apache.org/jira/browse/THRIFT-4806
 Project: Thrift
  Issue Type: Improvement
  Components: Node.js - Library, TypeScript - Library
Reporter: Brian Forbis


I'd like us to consider migrating the NodeJS library to typescript. I feel this 
would simplify the existing libraries and build system. The typescript library 
would just need to be compiled down to javascript as part of the build system.

This will:
 * Introduce compile time type checks to help catch bugs
 * Provide better type documentation for consumers of the library
 * Include better typescript tests in the build
 * Allow us to merge the nodets and nodejs lib folders
 * Synchronize the javascript codebase with the type definitions (currently, 
there is an external project at DefinitelyTyped to provide typescript types for 
the nodejs lib)

 

If the maintainers think this is a good idea, I'd be happy to take a stab at it.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)


[DISCUSSION] JIRA-Tickets vs. Patches & Pull Requests

2019-02-16 Thread Jens Geyer
Hi all,

a few days ago Jim raised the question about the policy regarding JIRA tickets.
As I am not fully happy with either solution we practiced so far, I gave it a 
second thought.

So here’s my proposal:

  1. We have one ticket per patch or pull request.
  2. Minor changes may be put into one patch/PR.

The first part should be obvious, so let me focus on the second point.

Larger changes should still be split into multiple Patches / PRs, because they 
usually come with their own language specific discussion, problems, 
implermentation details and last not least test cases.  In contrast, if a patch 
is quite minorish and affects a lot of languages needing a 3-liner fix in 
(basically) the same way, there is no point in creating 20 tickets just for the 
sake of having them.

A larger patch is defined by “changes that cannot easily be 
managed/commented/reviewed by one single person as a whole”. Yes, that’s a 
vague criterion, but it should only serve as a “intended outcome” kind of 
guide. The whole idea is to reduce unnecessary administrative overhead where 
possible, but split up matters where necessary. That’s what I want to achieve.

Could that work for us? Are there any objections? Other (maybe better) ideas?

Have fun,
JensG