Primary performance improvement wanted

[genyez]

I‘ve seen many improvable points, such as:

• don't copy data from/to string/vector by using emplace_back
• multi-threaded xml w/r
• multi-threaded com/decompression
• incremental save (base on dirty mark for example)

Any one interested in those?

Besides where is the original repo of zippy? I searched but end up with a nimble library. Any clue would be appreciated!

[Rozhdestvenskii]

Besides where is the original repo of zippy?

probably https://github.com/Shinmera/zippy

[genyez]

Besides where is the original repo of zippy?

probably https://github.com/Shinmera/zippy

Thanks for replying, but it seems to be a lisp repo, not a C one.

[troldal]

The Zippy library has been renamed KZip and can be found here:

https://github.com/troldal/KZip

It's basically just a wrapper around miniz:

https://github.com/richgel999/miniz

Regarding your suggestions for multithreading, that would be great features to have, but really depends on the underlying libraries (pugixml and miniz). None of them are currently multithreaded. With pugixml, reading the XML DOM tree can be done in parallel (provided it's not being altered at the same time), but writing is a different matter.