TE
TechEcho
Home24h TopNewestBestAskShowJobs
GitHubTwitter
Home

TechEcho

A tech news platform built with Next.js, providing global tech news and discussions.

GitHubTwitter

Home

HomeNewestBestAskShowJobs

Resources

HackerNews APIOriginal HackerNewsNext.js

© 2025 TechEcho. All rights reserved.

Contrastive Code Representation Learning: deep type prediction for TypeScript

7 pointsby parasjalmost 5 years ago

2 comments

pkmr103almost 5 years ago
1. Null Exceptions use for this tool? Looking at Splunk exception logs for a $4b Ecommerce company, a lot of stack traces ended up in Java null exception errors - very common for 25 year old ATG Java based code. So many in fact that tech debt was immense, waiting for years in both baseline and user code. This tool would be great application to detect without waiting for an actual exception. 2. What would be use of ML in E-commerce software where each cart checkout is often plagued with dozens of software errors, slowing down or killing the closing of the sale. So it all impacts final conversion from say 6% down to 3% in an ideal situation. On mobile it is worse where conversions are typically 0.5%. Of course other factors not just speed or software errors are at play. 3. Deep learning would help greatly in dynamic languages like JS,Typescript, even Python where compilers are limited in checking and lots of code errors fly under the radar.
ajayjainalmost 5 years ago
I'm one of the authors of this paper, feel free to ask questions.