Actions:
|
2017-06-08 07:57 AEST by Arthur Barrett - We could automatically alter documents on checkout. Actually this could apply to source code, web pages,
etc. but probably the most notable case is for documents.
Particularly we could alter whitespace.
If we understand the file format well enough (e.g.: Microsoft Office documents, Microsoft Word Documents
.doc/.docx), maybe even Postscript / PDF files, then we could get really clever and alter kerning.
Why?
The reason is security.
With a 'pattern' of particular spaces (or kerning) you can identify that that document was checked out by
person A (or maybe more accurately, came from session 123).
See this tweet for a little more info:
https://twitter.com/DisarmPSU/status/871903871882416128
Despite some googling, I wasn't able to find much documentation/discussion on the subject - but based on
the comments on twitter - it's fairly well understood.
On 'commit' we'd need to strip it all back out of course. Which makes this enhancement related to and/or
dependant on:
Bug 5978 : enh: keyword expansion for binary files
Bug 6218: enh: store docx, pptx, xlsx files unzipped with keywords (kkv)
We're really talking about storing some files with a special 'keyword' which means that we don't store the file
'as is' - but we pull it apart and when it is checked out we produce an 'edition' specific for that user.
e.g.: maybe this is behaviour that only gets used when the 'checkout' user has read-only access.
or maybe we allow some users to 'checkout' docx files as a PDF directly. |