- Alternative to #125: HTTP Authentication
- Raised by
- Ari Epstein?
- Date opened
- Saturday, 6 June 2009
- I wanted to suggest another way to implement support for pmwiki in environments where some pages are blocked from reading by HTTP/SSO authentication. The current workflow seems backwards to me:
1. wikipublisher sends the URL containing parameters to mkpdf.pl
2. mkpdf.pl fetches the URL from the wiki, having to somehow assume the user’s credentials — a problem especially if you want to share a pdf making server amongst several systems or use it over an insecure network
- Proposed solution
- This order of operations seems more logical to me:
1. wikipublisher posts the wikibook markup to mkpdf.pl
2. mkpdf.pl returns the pdf to wikipublisher which passes it back to the user
In my view mkpdf.pl should only be a backend service which either returns a pdf or one of a defined set of error codes that wikipublisher or another application can interpret and present to the user.
It would get somewhat more complicated if images also required the user’s credentials, although I think this could be addressed by including them as multipart attachments to the post in step 1 and having them keyed to references in the generate wikibook markup.
- This is an interesting proposal and it would be useful to see a working system based on this design. Maybe somebody would like to take on a project to remix the Publish PDF library so it works in the way proposed. It also creates the possibility of caching the generated PDF files, perhaps in the upload directory, so the reader has the option of downloading an existing PDF or generating a new one.
- Date closed
« 00136 · Edit Form · 00138 »