3 ms·
I implemented some options for that purpose in SingleFile [1]. They allow you to save the page when you bookmark it and eventually replace the URL of the page w
by gildas 6y ago
I implemented some options for that purpose in SingleFile [1]. They allow you to save the page when you bookmark it and eventually replace the URL of the page with the file URI on your disk.
[1] https://github.com/gildas-lormeau/SingleFile https://github.com/gildas-lormeau/SingleFile
- johnchristopher 6y agoCool, I was so into maff back in the days. I'll give it a try. (I even wrote this before checking out your link: Have you heard of https://en.wikipedia.org/wiki/Mozilla_Archive_Format https://en.wikipedia.org/wiki/Mozilla_Archive_Format from two or three Internets ago ? If so what's your thoughts on it ?)
- gildas 6y agoI would recommend you to take a look at SingleFileZ [1], it should remind you of something ;) [1] https://github.com/gildas-lormeau/SingleFileZ https://github.com/gildas-lormeau/SingleFileZ
- toomuchtodo 6y agoWhy a zip file instead of a WARC file? https://en.wikipedia.org/wiki/Web_ARChive https://en.wikipedia.org/wiki/Web_ARChive
- gildas 6y agoBecause it's easier to produce and extract. The zip format also allows creating self-extracting files (I'm referring to SingleFileZ). I'm not sure this is possible with the WARC format.
- toomuchtodo 6y agoI see you answered this in a thread a year ago [1] (came up in a Google search), my apologies. [1] https://news.ycombinator.com/item?id=21426056 https://news.ycombinator.com/item?id=21426056
- walski 6y agosee: https://github.com/c9fe/22120#why-not-warc-or-another-format-like-mhtml- https://github.com/c9fe/22120#why-not-warc-or-another-format... > Both WARC and MHTML require mutilatious modifications of the resources so that the resources can be "forced to fit" the format. At 22120, we believe this is not required
- gojomo 6y agoThis is false. If you're doing WARC correctly, HTTP resources/responses are stored verbatim. Perhaps the only possible purer format would be packet captures, say of the full HTTPS session, along with the session keymatter and connection metadata to later extract the verbatim HTTP resources. That'd be interesting, but I doubt that's what this "22120 format" (for which I see no documentation links) does.