[recoll-user] Re: Something special / unusual about .cpp suffix?

  • From: jfd@xxxxxxxxxx
  • To: recoll-user@xxxxxxxxxxxxx
  • Date: Tue, 20 Sep 2011 17:15:23 +0200

Randy Kramer writes:
 > For quite some time, I've been using recoll to index and search the source 
 > code for scintilla and SciTE.  Seems to work fine.
 > 
 > Recently, I wanted to search the recoll source code and tried to set things 
 > up 
 > the same way.
 > 
 > The one thing I had to do (in both cases) was set up a mimemap file--in the 
 > case of scintilla/SciTE, it contains:
 > 
 > .h   = text/plain
 > .c   = text/plain
 > .cxx = text/plain
 > 
 > For recoll, it contains
 > 
 > .h   = text/plain
 > .c   = text/plain
 > .cxx = text/plain
 > .cpp = text/plain
 > 
 > Although I could be doing something dumb, I've tried this quite a few times. 
 >  
 > Searches for all kinds of things (e.g., #include statements) work fine for 
 > scintilla/SciTE, but don't work for recoll.
 > 
 > Any hints appreciated!
 > 
 > (I have a recoll.conf file in both cases that specifies topdirs (only).  I 
 > run 
 > recoll and recollindex from the parent directory of the source code tree in 
 > both cases.  If I search for, for example, a file name (e.g., mh_mbox), 
 > recoll shows mh_mbox.cpp in the search results, but if I search for stuff 
 > within mh_mbox that I know is there, I get either "No results found" or the 
 > list of results contains no .cpp files.)
 > 
 > Randy Kramer


Mmmouark ark! ark! ark! (sardonic laugh) Recoll text is unsearchable, that's
part of the plan for word domination.

More seriously, set up the recoll.conf that you use for indexing the recoll
source so that you get a detailed log:
  logfilename = /tmp/mylogforrecoll
  loglevel = 6

Then perform a full indexing pass:
recollindex -z -c /path/to/the/confdir/for/indexing/recoll

And look in the file to see what happens to .cpp files (to make things
easier, search for a file whose name won't be listed as a message origin,
like for example ssearch_w.cpp).

If nothing obvious appears, please send me the log fragment for one of these
files. Maybe I'll see something, else we'll check how you perform the
queries too.

By the way, with recent versions (at least since 1.14), you should not have
to change mimemap all that much, .c, .cpp, .cc, and .h files are supposedly
indexed as text (I just added .cxx).

Cheers,

jf

Other related posts: