What does bufferedWrite do?

[gera2ld]

Could you please explain what bufferedWrite exactly does?
I tried writer.add with/without bufferedWrite but it seems to make no difference, either in output or performance. I'm also wondering if it's safe to add multiple files in parallel with the default option bufferedWrite: false.

[gildas-lormeau]

When bufferedWrite is set to true, zip.js creates a separate buffer for each entry being created to avoid data corruption in the zip file. However, when it's set to false, zip.js can still detect if ZipWriter#add is called in parallel and create automatically the buffers if necessary. In this case, the entry added via the first call to ZipWriter#add is blocking because zip.js did not know there would be concurrent writes. So the performance depends on the size of the first entry added in the zip. That's the reason why you should set bufferedWrite to true when you know you will call ZipWriter#add multiple times in parallel. Any value is safe though.
Finally, note that the option keepOrder is also set to true by default. It's necessary when writing files that could rely on zip64. Since entries are ordered, this can have a similar impact on performance. If you know that the zip file does not need zip64 (e.g. if it weights less than 4GB), then you can set keepOrder to false. Thus, there won't be any blocking lock when adding entries in parallel.

[gera2ld]

Thanks a lot for the detailed clarification, that's very clear. I guess it would be great to have this in the documentation.

[gildas-lormeau]

I agree this should be documented. Meanwhile, I created a discussion thread here: https://github.com/gildas-lormeau/zip.js/discussions/237 in order to keep it visible.