Re: How Google indexed a file with no external link

[Kevin <kevin () brasscannon net>]

The problem (at least in my case) has been resolved, so if you've been 
checking my site you can stop now:   ;-)

> >     http://handsonhowto.com/?M=A
> >     http://handsonhowto.com/?S=D

I now get the index.html page, as intended.

The fix was to take all the "Indexes" options out of my httpd.conf; all 
of my docroots have index.html files (including one for the bare IP 
address*), so I won't miss having it turned on.

*(This covers the case where someone hits the IP address instead of a
named virtual host; rather than defaulting to one page out of many, I 
put up a "You are not here" page listing the named sites.)

This does appear to be an Apache bug, and now that we have a handle on
it, I'll report it as such.

The prize for the most code-oriented response:

> Looking at the Apache source: src/modules/standard/mod_autoindex.c 
> we see:
> -----------------------------< cut here >-------------------------------
> /*
>  * Define keys for sorting.
>  */
> #define K_NAME 'N'              /* Sort by file name (default) */
> #define K_LAST_MOD 'M'          /* Last modification date */
> #define K_SIZE 'S'              /* Size (absolute, not as displayed) */
> #define K_DESC 'D'              /* Description */
>
> #define D_ASCENDING 'A'
> #define D_DESCENDING 'D'
> -----------------------------< cut here >-------------------------------
>
> It looks as though the auto-indexing ignores the existence of index.html
> if provided with params.  This seems to be the QUERY_STRING handling in
> index_directory().
>
> I don't know Apache well enough, but I would suggest that the bit of the
> logic which handles "/" -> "/index.html" comes earlier in the parsing,
> before deciding to hand things off to mod_autoindex and that this first
> rewriting logic can't handle query-strings.

Suggestions about a quick-fix in httpd.conf were received from many 
parties, and it worked immediately.  My thanks to all.