Parcel: Publish benchmarks app

[jsonnull]

I'd love for the app the README benchmarks are based on to be published, so that these can be analyzed and improved.

Personally, any time I see benchmarks I want to see:

• How can I reproduce?
• How many runs are the benchmarks based on?
• Was best run or average run considered?
• What were the configurations? Caching in the parcel config was mentioned, but what about caching capabilities of webpack, babel, etc...?

[devongovett]

Unfortunately, I can't publish that app as it is closed source. Do you know of a reasonably large open-source app that we could try benchmarking against?

[elrumordelaluz]

Probably not huge, but why not one from here https://github.com/gothinkster/realworld

[benderTheCrime]

Happy to create a separate issue as well, but I'd also like to see comparisons against Rollup and some other bundlers!

[azu]

I'm interested in the source-map setting of the benchmark app(webpack and browserify).
Because, parcel-bundler does not support source-map yet #68
We have known that generating source-map is very slow.

[jsonnull]

@DeMoorJasper why was this issue closed?

The pull request which referenced this issue swaps out the JSON parser used internally by parcel. It does not include any performance benchmarks or notes related to performance.

[DeMoorJasper]

There are various examples and blog posts showing that parcel is faster than webpack.
The big project referred to in the readme is private, so thought it would be safe to close this old issue.
If you want me to reopen I will.

[jsonnull]

Every question in this thread aside from "can you publish the benchmark app from the readme" has not been answered.

Since the benchmarks in the README are the most visible to users of the project, it seems reasonable to use a public app to ensure that a fair comparison was made.

Now that parcel is older, in addition to the above unanswered questions I would like to know:

• Which versions of the projects are being compared?
• Which features of those projects are used? It should be an apples-to-apples comparison.

[mpalmr]

This looks really shifty and although Parcel looks like good tech, this particular issue really taints it in my mind. Why not just write another benchmark? I wouldn't mind doing so myself. When I look at benchmarks like this it jus makes me think the authors of Parcel are being straight up disingenuous.