On Sat, Aug 22, 2015 at 9:17 AM, ArtGiray . <giraybal(a)gmail.com> wrote:
They're almost the same, they wind up routing to the same includes/actions/
code. Clients may treat the two requests differently because a request to
index.php request has an inferred "PHP" content-type, but for a bot it
shouldn't matter. (There is an RFC to switch WMF sites to the former style
of action URL [1], but both will continue to work.)
When I use `curl --dump-header -` with them they produce identical HTML,
but the HTTP headers are different, the first outputs a
Transfer-Encoding: chunked
and the latter
Content-Length: 58015
These are probably artifacts of the Varnish caching that WMF sites use, but
it's interesting.
(i used first one before, then use second, but sometimes it's not
response...)
I've never heard of that, but the first time I tried the index.php request
using curl after about 20 seconds the connection closed with no response!
Is that what you experienced? You or I should file a bug.
are any of these queries doing extra processing on
server side ?
I believe not when it gets to a PHP server, but something in WMF's complex
configuration is treating them differently.
which one i should use for more speed on my wikibot?
I don't know, sorry.
(or any different query suggestion?)
Read [2] and [3]. Depending on what you want to do with wikitext there are
other APIs.
[1]
https://www.mediawiki.org/wiki/Requests_for_comment/Clean_up_URLs
[2]
https://www.mediawiki.org/wiki/API:FAQ#get_the_content_of_a_page_.28wikitex…
[3]
https://www.mediawiki.org/wiki/Manual:Parameters_to_index.php#Raw
--
=S Page WMF Tech writer