Update of /cvsroot/archive-access/archive-access/projects/wayback
In directory sc8-pr-cvs1.sourceforge.net:/tmp/cvs-serv8448
Added Files:
plan.txt
Log Message:
* plan.txt
project roadmap/todo/brainstorming notes; initial commit. extend/revise at will
--- NEW FILE: plan.txt ---
== NGWM NEXT STEPS ==
0.2 (week of Nov 28 - Dec 2)
- fix windows file-pipeline bug (move pipeline state into bdb?)
- better install/admin instructions (in file and in UI)
- bundle small default dataset (sample ARC) for immediate use
- remove extraneous files from build distro bundle
- polish UI to clarify local operation/collection/etc.
- release via SF 'file release' mechanism, announce to lists
- demo for team next Tuesday (Dec 6) for UI/feature feedback
0.4 (by first week of January)
- verified performance at typical scale of contract collections
- retrieve/cache/respect freshest robots.txt
- handle multiple named collections
- nice, flexible results-list UI (classic calendar or other)
- admin tasks password-protected
- nice to have: manual exclusions
- nice to have: floating in-page indicator of WM date/status
- nice to have: clean nutch integration
0.6 (as necessary)
- TBD
== BRAINSTORMING ==
Error-handling:
- when URL not available
- if auto-fetch, make sure an in-page floaty announces that it's
a fresh retrieve; log to side for QA purposes
- if not auto-fetch, offer link to broader search (all local collections)
or remote collections (such as public wayback machine)
Entry pages:
- when scanning ARCs, offer option to add /root pages to 'entry pages' list
- also allow admin to add entry pages
- display recommended entry pages below main search box (or on separate page)
Search/Admin UI:
- make host, collection very clear in UI (unless suppressed)
- highlight especially 'localhost' connections (different color background?)
- ape google/search-engine style to maximum extent practical
Replay UI:
- in-page presence
- collapsable uri/date/collection indicator
- mouseover indicators?
- n/a graphic for n/a images
- stayback firefox extension
- would look for special indicator inside page that it's a wayback session
- if found, would refuse to load inline resources or follow clicks without
patching URI back to wayback machine (assuming intercept/callback is
possible)
Heritrix integration ideas:
- bundled? just installed alongside?
- simultaneous update during crawl?
- 'not here yet but scheduled' error/placeholder IMGs?
- schedule-when-requested-via-WM?
- linkify all logs/reports to WM?
|
