Hacker News new | past | comments | ask | show | jobs | submit login

This reminds me of a precompressed template system I built for an old Python Tornado webserver.

The system precompressed the static parts of the template, with matches only self-referencing inside each span of static content.

The dynamic parts of the template were prefixed with 'literal' tags.

This meant that serving a gzipped stream was as simple as concatenation, just as it was for normal uncompressed templates.

It could have been a lot cleverer - it could have huffman'ed the dynamic parts, for example - but it was an easy win.

Later, our Tornado servers were migrated behind a caching proxy and the whole purpose of serving up (partially) compressed pages vanished.




Done that too. Kind of like userland distributed vectored IO.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: