(no commit message)

[ikiwiki.git] / doc / bugs / trouble_with_base_in_search.mdwn

diff --git a/doc/bugs/trouble_with_base_in_search.mdwn b/doc/bugs/trouble_with_base_in_search.mdwn
index 4ca2b4beff9903ecfde77b709cd25729c83cda5a..7c3d133872ebdca6c25de35657764f3abd3dc3a5 100644 (file)
--- a/doc/bugs/trouble_with_base_in_search.mdwn
+++ b/doc/bugs/trouble_with_base_in_search.mdwn
@@ -51,3 +51,21 @@ And I'm sure someone else could come up with something better and more general.
 > directly, my HTTP vs. HTTPS cleanup in recent versions should mean that
 > you rarely get redirected to `alfred`, because most URLs are either
 > relative or "local" (start with '/'). --[[smcv]]
+
+>> I did try setting `url` to the "Alfred" machine, but that doesn't seem clean to me at all, since it forces someone to go to Alfred when they started off on Betty.
+>> Even worse, it prevents me from setting up a test environment on, say, Cassandra, because as soon as one tries to search, one goes to Alfred, then Betty, and not back to Cassandra at all.
+>> Hardcoded solutions make me nervous.
+
+>> I suppose what I would like would be to not need to use a `<base href>` in searching at all.
+>> --[[KathrynAndersen]]
+
+>>> `<base href>` is *not* required to be absolute in HTML5, so when
+>>> `html5: 1` is used, I've changed it to be host-relative in most cases.
+>>> I think that at least partially addresses this bug report,
+>>> particularly if we [[todo/generate HTML5 by default]] like I've suggested.
+>>>
+>>> The `<base>` is there so we can avoid having to compute how to
+>>> get to (the virtual directory containing) the root of the wiki from
+>>> `ikiwiki.cgi`, which might well be somewhere odd like `/cgi-bin/`.
+>>> I think there are probably other things that it fixes or simplifies.
+>>> --[[smcv]]