What is the reason of using a sync version of writeFile for pdf save?

[tsabolov]

https://github.com/GoogleChrome/puppeteer/blob/9b0a06216e5b364aecaef6931d7844a91bebe46e/lib/Page.js#L588
Why wasn't async version of writeFile used for writing the pdf?

[paulirish]

I presume there isn't a strong reason.. :)

If you wanted to change it, I'm sure the eng team would appreciate a pull request.

[Garbee]

If you do use the async version, it is still going to be awaited for processing. So is there any functional difference between the two at that juncture from the point of view of code operation?

[paulirish]

@Garbee reading lib/fs and this post it seems like the big difference is that the node event loop is blocked with sync, so any events coming from the chrome child_process (for example) would not be recognized until after.

And there's this issue suggesting that sync will always have inferior performance characteristics.

i can't think of a reason we NEED sync here, so might as well do the right thing.

[Garbee]

big difference is that the node event loop is blocked with sync

SGTM then.

[paulirish]

Ah. there's already a PR for this: #879