Talk:Offline content generator/Bundle format

Few questions

I have a few technical question about the intermediate storage solution:

• How will be stored this Bundle? This will be a directory on the fs, a ZIP file, ...?
• Why we need "html.db" and "parsoid.db", I thought parsoid output is enough? Traditional API HTML output does not allow an efficient HTML rewriting.
• Why storing HTML output in a sqlite db? This is slower than the fs, I don't see the added value over the fs... and worth: this doesn't scale correctly.
• From where come the book metadata (for example the list of articles)? From the Front-end REST request or from the render client challenging the Collection extension (that means that this would only work with the collection extension)?
• Why do we need imageinfo.db, the HTML output already knows which file it needs to be correctly rendered?
• Where would be the problem if we would avoid any intermediate storage for the HTML/medias and write "directly" the parsoid output to the end-format?

Kelson (talk) 16:35, 19 November 2013 (UTC)

It's stored in a zip file, although "directory on the fs" is more or less equivalent.

Note that this was originally documenting the format currently used by PediaPress's mwlib, so it contains things that they use and we won't, possibly things they have in there for debugging, and also possibly things that they used to use but don't anymore. Then it was decided to extend this format, and we haven't made the decision yet if we can get rid of the bits we don't need but PediaPress might. Anomie (talk) 13:16, 20 November 2013 (UTC)