Subject: Re: [htdig] 3.2.0b2 problems (was: Local digging files without extension)
From: Markus Warg (firstname.lastname@example.org)
Date: Wed Oct 18 2000 - 02:29:45 PDT
On Die, 17 Okt 2000, Gilles Detillieux wrote:
> Well, I don't see the connection between Marcel's email and my reply to
> him, which you quoted below, and your problems with compiling 3.2.0b2.
> I don't think 3.2's handling of local files and the valid_extensions
> attribute is any different than 3.1.5's.
Not a direct connection, indeed. But as I stated, I came into the game
when Marcel encountered those core dumps. So our current problem
just switched to another level :)
> 3.2 is still in beta releases, so it's still very rough around the edges.
> However, if you're taking up Geoff's challenge to develop a patch to
> get Marcel's requested features in there, then 3.2 may be the way to go.
I think Marcel has the one or the other proposal for mime type detection,
I just added a quick hack that would fit for our needs only. I just
set the file extension to ".txt" if there is no extension. This is quite
sufficient for us because there are around 300000-400000 files without
extension, all contain plain text. There are no other files around.
Of course this is nothing to add to a release version.
> The problem you've run into is well-known, but unfortunately the solution
> is eluding us at the moment. For some reason, when using shared libraries
> on Solaris, objects in this library are not initialized properly.
> The only workaround right now is not to use shared library support on
> Solaris, by using the --disable-shared option on ./configure.
Okay, I'll try this out. I couldn't believe this first time I saw
this behavior. Mhh, I built some dynamic libs for OS/2 some time ago,
if I remember right, there is a feature that calls initialization code
automatically when a module gets loaded. Maybe this fails for Solaris?
> As for fixes to b2, there are plenty, but not in the form of patches.
> You should grab the latest development snapshot for 3.2.0b3, which is
> still in the works, but probably more stable right now than b2 is.
Okay, for now I'll try b2, when our tests are finished I will
get the new source from cvs (I haven't done this before and I just
don't have the time right now to get this running).
Thx for the information.
> According to Markus Warg:
> > Gilles,
> > I just got that forwarded from Marcel, we upgraded from 3.1.x to the
> > current version 3.2.0b2. Now there is a problem with that version,
> > I don't know if this is really the latest development version. If yes,
> > I've a big problem when compiling on a sun solaris 2.7 with gcc 2.95.2.
> > Everything goes fine until we start htdig. We get an arithmetic exception
> > directly after starting htdig. I ran the program with gdb as it was
> > fortunately not stripped. Here is the backtrace (I added some debug code
> > to Dictionary.cc, so line numbers may vary by one or two):
> > #0 0xff036dac in .urem () from /usr/lib/libc.so.1
> > #1 0xff1759bc in Dictionary::Add (this=0x9d9ec, name=@0xffbef3f0, obj=0x9f278)
> > at Dictionary.cc:199
> > #2 0xff174134 in Configuration::AddParsed (this=0x9d9ec, name=@0xffbef3f0,
> > value=@0x9f278) at Configuration.cc:197
> > #3 0xff174d50 in Configuration::Defaults (this=0x9d9e8, array=0x9c230)
> > at Configuration.cc:378
> > #4 0x309e0 in main (ac=4, av=0xffbefbdc) at htdig.cc:122
> > Line 199 in unmodified source is around line 197:
> > int index = hash % tableLength;
> > What brought me to the idea that this value (tableLength)is 0. I added
> > debug code to all four constructors and found that NONE OF THEM was
> > called before the function Add was called. I see the call to the
> > constructor in Configuration.h (like Configuration:<variable>(value),...).
> > Do you have any idea why there is no call to a constructor for Dictionary?
> > I'm quite new to htdig, Marcel is out of my reach for some hours, so
> > please excuse me if there are already fixes or patches to beta 2.
> > best regards,
> > Markus Warg
> > On Die, 17 Okt 2000, Marcel Hicking wrote:
> > > Oweh.
> > >
> > > ------- Forwarded message follows -------
> > > From: Gilles Detillieux <email@example.com>
> > > To: firstname.lastname@example.org
> > > Date sent: Mon, 16 Oct 2000 11:56:24 -0500 (CDT)
> > > Copies to: email@example.com
> > > Subject: Re: [htdig] Local digging files without extension
> > >
> > > According to Marcel Hicking:
> > > > I'm trying to dig local files using a generated start_url
> > > > list and local_url.
> > > >
> > > > The files to index are not accessible via http but since
> > > > I'm going to pipe the search results through a wrapper
> > > > anyway they won't cause any problems.
> > > >
> > > > So far the start_url/local_url stuff works fine but
> > > > htdig -vvv says:
> > > > ...Trying local files
> > > > found existing file /foo/bar
> > > > Local retrieval failed, trying HTTP
> > > >
> > > > User righs are alright (I can cat the files ;-)
> > > >
> > > > I suspect the problem to be the missing file
> > > > extension and the therefore unkown mimetype.
> > > >
> > > > Any chance of setting some sort of default mimetype?
> > > > Any other workaround?
> > >
> > > Yes, the problem is the missing extension. No, there's no
> > > config attribute you can set to assign a default mime type
> > > to files with no extension. Yes, there is a workaround,
> > > which would be to change the RetrieveLocal() method in
> > > htdig/Document.cc to handle this case and assign the type
> > > you want. We don't do this in the distributed source
> > > because there isn't universal agreement on what mime type
> > > these files should have, and we haven't worked out a better,
> > > more configurable scheme for this code yet.
> > >
> > > --
> > > Gilles R. Detillieux E-mail:
> > > <firstname.lastname@example.org> Spinal Cord Research Centre
> > > WWW: http://www.scrc.umanitoba.ca/~grdetil Dept.
> > > Physiology, U. of Manitoba Phone: (204)789-3766 Winnipeg,
> > > MB R3E 3J7 (Canada) Fax: (204)789-3930
> > >
> > > ------------------------------------
> > > To unsubscribe from the htdig mailing list, send a message
> > > to email@example.com You will receive a message to
> > > confirm this. List archives:
> > > <http://www.htdig.org/mail/menu.html> FAQ:
> > > <http://www.htdig.org/FAQ.html>
> > >
> > > ------- End of forwarded message -------
> > > --
> > > VIA NET.WORKS Deutschland GmbH http://www.via-net-works.de
> > > Bismarckstrasse 120 fon +49 203 3093-101
> > > D-47057 Duisburg fax +49 203 3093-112
> > > Deutsche Provider Network firstname.lastname@example.org
> > --
> > VIA NET.WORKS Deutschland GmbH email@example.com
> > Bismarckstr. 120 www.via-net-works.de fon: +49 203 3093-101
> > D-47057 Duisburg Deutsches Provider Network fax: +49 203 3093-112
> Gilles R. Detillieux E-mail: <firstname.lastname@example.org>
> Spinal Cord Research Centre WWW: http://www.scrc.umanitoba.ca/~grdetil
> Dept. Physiology, U. of Manitoba Phone: (204)789-3766
> Winnipeg, MB R3E 3J7 (Canada) Fax: (204)789-3930
-- VIA NET.WORKS Deutschland GmbH email@example.com Bismarckstr. 120 www.via-net-works.de fon: +49 203 3093-101 D-47057 Duisburg Deutsches Provider Network fax: +49 203 3093-112
------------------------------------ To unsubscribe from the htdig mailing list, send a message to firstname.lastname@example.org You will receive a message to confirm this. List archives: <http://www.htdig.org/mail/menu.html> FAQ: <http://www.htdig.org/FAQ.html>
This archive was generated by hypermail 2b28 : Wed Oct 18 2000 - 02:34:55 PDT