Different search results depending on whether <EXEC> is used to start the search.

barry.marcus
Posts: 288
Joined: Thu Nov 16, 2006 1:05 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by barry.marcus »

I compromised by going back to the way it was and calling the *separate* script using <EXEC> rather than <EXEC BKGND>. Again, the results are the same. The script, when called with <EXEC> returns different results, even with the BKGND option.

(Granted, this may not be a definitive test, since it's back to being two separate scripts. But they look to be identical.)
User avatar
mark
Site Admin
Posts: 5513
Joined: Tue Apr 25, 2000 6:56 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by mark »

One thought. You're not execing a different version of texis than you're running in the webserver are you?

Test by adding to both flavors:

{Texis version: <vxinfo version>$ret}
barry.marcus
Posts: 288
Joined: Thu Nov 16, 2006 1:05 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by barry.marcus »

Everything is all on the same server, and there is only one version of Texis installed on it, so no... the versions are the same.

As for which which result set is correct... Well, there seems to be issues with *both* result sets, frankly. I suspect that this stems from whatever issue is causing the anomalous results described in [Tkt#4ea85e0ab]. The query here has all the same characteristics as those described in that open issue. (It's crossed my mind that all of this is part and parcel of the same underlying issue, but I can't say for sure.) The foreground search returns 2195 hits while the background search returns 2199 hits. Moreover, ALL of the hits from the foreground search are included in the background search results. That is, there are four distinct additional hits returned by the background search. But here's where things get a little weird... Of those four, three look like they should NOT be hits. One of them, however, looks to be a legitimate hit, and yet is NOT returned by the foreground search. As I said before, the queries are big and complex, the fields being searched contain a lot of data per record, and there are many records. Which is all to say that we don't yet know if there other problems with either result set.

Argh!
barry.marcus
Posts: 288
Joined: Thu Nov 16, 2006 1:05 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by barry.marcus »

HOLD THE PRESSES!!!! The version numbers ARE different!?! I apologize, but I am really confused how that can be!!! I don't understand.

Foreground:
2195 results
Commercial Version 6.00.1289279282 20101109 (i686-intel-winnt-64-32)

Background (without BKGND):
2199 results
Commercial Version 6.00.1317406931 20110930 (i686-intel-winnt-64-32)


Background (with BKGND):
2199 results
Commercial Version 6.00.1317406931 20110930 (i686-intel-winnt-64-32)
User avatar
jason112
Site Admin
Posts: 347
Joined: Tue Oct 26, 2004 5:35 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by jason112 »

A CGI texis.exe lives in your web server's CGI directory, which might have been moved around since the original install and would cause an upgrade to not change that file. You should be able to copy your installationdir's texis.exe over the CGI one to bring them both back in line.
barry.marcus
Posts: 288
Joined: Thu Nov 16, 2006 1:05 pm

Different search results depending on whether <EXEC> is used to start the search.

Post by barry.marcus »

Mystery solved. So, to all who helped me on this... Yes, my mistake, and I apologize for all head-scratching this caused ... Some time back, in an effort to streamline our environment, I changed the CGI directory to something other than the default. During an upgrade of Texis, however, I suppose failed to indicate to the installer what that new directory was, and from that point on we've been running different versions, depending on how Texis is called.

Thank you, as usual, for your efforts.
Post Reply