On 12/25/05, Magnus Manske <magnus.manske(a)web.de> wrote:
For practical purposes, we should consider offereing a
dump that contains
* articles
* templates (which are needed for many articles)
* a list of all author names
That would contain everything needed to generate GFDL-compliant,
complete articles, but still be much smaller than the full dumps.
We could put this information in the "articles" dump (slightly larger
download for everyone) or create a new dump type (more memory and
processor power needed to create and offer it).
That is likely not enough, because you couldn't discover the principal
authors that way... a great many republication worthy articles are
almost entirely written by one or two people. But more importantly, a
mere list of names would not appear to be enough to satisify the
GFDL's requirement to 'Preserve the section Entitled "History"'.
It would be preferable to only include templates which are used in the
main namespace, although that would further complicate the dumping
process.