![]() |
SEARCH'97 Information Server V3.61 Release Notes |
This document contains the latest information for SEARCH’97 Information Server V3.61. The following topics are covered:
This release includes bug fixes and the Verity Query Language Quick Reference in
PDF format. You can print a copy for personal use from /prodnote/vql_ref.pdf on
the CDROM, and from /installdir/s97is/locale/english/doc/prodnote/vql_ref.pdf after
you have installed Information Server V3.61.
SEARCH’97 Information Server V3.61 and SEARCH’97 Agent Server V1.12 are now supported on the following platforms:
| Platform Directory | Platform Description |
| _nti31 | Microsoft Windows NT Server 4.0 (SP3) |
| _solaris | Sun Solaris 2.5, 2.51 |
| _aosf40 | DEC Alpha UNIX 4.0 |
| _hpux10 | Hewlett-Packard HP/UX 10.10, 10.20 |
| _irix62 | SGI IRIX 6.2, 6.3 |
| _rs6k42 | IBM RS/6000 AIX 4.13 and 4.2 |
If you are installing SEARCH’97 Information Server V3.61 on HP-UX from
a locally mounted CDROM, the following information pertains to the UNIX mount
command.
HP-UX converts all the files on a CD to uppercase and appends ;1 to the end
of each file. To mount a CD on HP-UX 10.x without the conversion, use the following
command-line switches for the mount command:
%mount -F cdfs -o cdcase /dev/dsk/c0t2d0 /cdrom
Your paths may differ for the CD system device and the final device name.
The KeyView feature list table below identifies supported formats and filter names. For each format, information about meta data extraction is provided. These keys are used in the feature list table:
|
Format |
Version |
Filter Name |
Summary Info |
Character Set |
|
ASCII text |
All |
afsr |
NO |
NO |
|
Applix Spreadsheet |
assr |
NO |
NO |
|
|
Applix Asterix |
V2.0 |
awsr |
NO |
NO |
|
MS Excel cvsr format |
csvsr |
NO |
NO |
|
|
DCA-RTF |
SC23-0758-1 |
dcasr |
NO |
YES |
|
IBM DisplayWrite |
V1.0, v1.1 |
dw4sr |
NO |
YES |
|
HTML |
All |
htmlsr |
PARTIAL |
YES |
|
Lotus 1-2-3 |
’96, ‘97 |
l123sr |
PARTIAL |
YES |
|
Lotus AMI Pro |
V2.0, v3.0 |
lasr |
PARTIAL |
YES |
|
Lotus Word Pro (NT only) |
’96, ‘97 |
lwpsr |
PARTIAL |
YES |
|
MS Word for Mac |
V4.x, 6.x |
mbsr |
NO |
NO |
|
MS Word for Mac ‘98 |
‘98 |
mbsr |
YES |
NO |
|
Adobe FrameMaker MIF |
V5.5 |
mifsr |
NO |
YES |
|
MS Word |
V1.x, v2.0 |
misr |
NO |
NO |
|
MS Microsoft Works |
V3.0, v4.0 |
mswsr |
NO |
NO |
|
MS Word |
V6.0, v7.0 |
mw6sr |
YES |
YES |
|
MS Word |
V8.0 |
mw8sr |
YES |
YES |
|
MS Word for DOS |
V2.2, to v5.0 |
mwsr |
NO |
NO |
|
MS Works Spreadsheet |
mwssr |
NO |
YES |
|
|
MS PowerPoint PC |
V4.0 |
ppcsr |
PARTIAL |
NO |
|
MS PowerPoint |
’95, ‘97 |
pptsr |
PARTIAL |
YES |
|
Lotus Freelance |
’96, ‘97 |
przsr |
NO |
NO |
|
Corel QuattroPro |
V7.0, v8.0 |
qpssr |
PARTIAL |
YES |
|
MS Rich Text Format |
rtfsr |
PARTIAL |
YES |
|
|
Corel Presentations |
V7.0, v8.0 |
shwsr |
YES |
NO |
|
UNICODE |
unisr |
NO |
YES |
|
|
Lotus 1-2-3 |
V2, v3, v4 |
wkssr |
NO |
YES |
|
WordPerfect for DOS |
V5.0, v6.0 |
wosr |
NO |
YES |
|
WordPerfect for Windows |
V7.0 |
wp6sr |
PARTIAL |
YES |
|
WordPerfect for Mac |
V2.0, v3.0 |
wpmsr |
NO |
YES |
|
MS Excel |
V3, 4, 5, ’97, ‘98 |
xlssr |
YES |
YES |
|
XyWrite |
V4 |
xywsr |
NO |
NO |
Lotus Word Pro filter is supported on Windows NT only. This filter is not supported on DEC Alpha NT or any UNIX platforms.
WordPerfect filters do not handle the n superior character properly. The current character mapping functionality for these filters does not support the display of the n superior character.
The Information Server Product Notes V3.6 for Solaris and Windows NT Server 4.0 only mentioned the following as a known problem:
“The current release of the KeyView Filter Kit is unable to index RTF documents generated by Oracle applications. The Verity indexing engine skips the documents.”
This is no longer an issue. You should be able to index such RTF documents.
If you are running Information Server V3.61 with Netscape Enterprise Server V3.51, you may encounter problems running the Information Server CGI scripts which use the POST request method if you click on cancel to stop a request before it is completed.
Applying patch B and patch C for Enterprise Server V3.51 fixes the problem. The patch files can be found at
http://help.netscape.com/filelib.htm.
On Solaris 2.51 with some patch levels, you may encounter problems running Information Server if you installed it as root. When you install as
root, ownership permissions of several Information Server files and directories are restored to the user account which created the tarfile,
rather than set to root. This problem only occurs when install.sh is executed by root, and not when you install
under any other user account.
After you have installed Information Server, you must set the permissions of all of the Information Server files to the same user account under
which your web server runs. Note that ownership is tied to the user account which runs the web server and not the owner of the actual web server
files. For example, root may own the files, but the web server daemon runs as nobody. Therefore, the Information Server
files should be owned by nobody.
When Verity Spider indexes documents into a collection, it populates a “Vspider Starts” value field in a nnnnnnnn.abt
file in the /assists directory of the collection. The .abt files are maintained for internal use and they contain information
about the collection. Each time Verity Spider is run, the “nnnnnnnn” part of the file name is incremented to create a new file,
and information is appended to the fields defined.
The “Vspider Starts” value field contains all of the accumulated starting points for all Verity Spider indexing jobs run for the
collection. When running refresh and restart jobs, Verity Spider uses this information to try to figure out the default options.
Following is a section of a “.abt” file:
field_name: "Vspider Starts"
{
value: "http://colt/Messages,http://colt:8015/rtfm/Messages,"
}
The “Vspider Starts” value entry can only contain 32000 characters. When this field exceeds 32000 characters, which it may do
after numerous indexing tasks and especially when the starting points are long, you will see an error message similar to the following.
mkvdk - Verity, Inc. Version 2.4.0 (_solaris, Dec 4 1998)
Error E0-1320 (Cdb): Argument string too long (max is 32000)
Error E0-1316 (Cdb): Error parsing
In this case, the message was produced by mkvdk. You will see a similar message when using Verity Spider to run an indexing job.
To remedy the error and thus allow indexing, you must open the “.abt” file which contains the overlong “Vspider Starts” value in
a text editor and clear the contents of the value: “startingpoints” line so that only value: remains.
NOTE: Always backup a file before you edit it. If anything goes wrong the next time you attempt to index, you can always copy over the backed up file and try the edit again.
In the above example error message produced by mkvdk, you can see the exact file which needs to be edited to clear the
“Vspider Starts” value field. If you are not sure which file to edit, choose the highest incremented filename. For example, choose 00000003.abt
instead of 00000002.abt or 00000001.abt.
If you are installing Information Server V3.61 with Netscape Enterprise Server V3.51 or Microsoft Internet Information Server V4.0, you must edit the DisplayName Registry key before you can install the Verity Connector for Lotus Domino/Notes V1.01.
The Registry key,
HKEY_LOCAL_MACHINE/SOFTWARE/Verity/Internet/3.00/DisplayName
will contain values for Enterprise Server V3.51 and Internet Information Server V4.0 which are unrecognizable by Verity Connector for Lotus
Domino/Notes V1.01.
For Netscape Enterprise Server V3.51, do the following.
For Microsoft Internet Information Server V4.0, do the following.
You may not be able to index files with unknown extensions with Verity Spider. Some examples, are
newacct.sam, lotustst.log, sample.wk.4, and filename.pre.
A workaround is to include the -mimeinclude option with a wildcard argument, “*/*.” For example:
vspider -collection name -start c:\testdocs -mimeinclude “*/*”
Note that the wildcard argument will index every file found. Use with discretion.
Your ability to index the local machine on which Information Server is installed depends on how the local machine is defined and how you refer to it. The following table depicts when the local machine can be indexed.
| If the local machine is defined as... | And you refer to it as... | Then... |
localhost anyIPaddressin hosts file. |
localhost |
You cannot index the local machine because of the keyword localhost. |
anyhostname 127.0.0.1in hosts file, where anyhostname
is not localhost. |
anyhostname or 127.0.0.1 |
You can index the local machine. |
anyhostname anyIPaddressin hosts file, where anyhostname is not localhost, and anyIPaddress
is not 127.0.0.1. |
anyhostname or anyIPaddress other than 127.0.0.1 |
You cannot index the local machine: a) if you are licensed for local host spidering only. b) if you are only using a hosts file. You can index the local machine if you licensed Verity Spider for remote spidering and your machine is defined as anyhostname in a DNS server. Whether or not you are using a hosts file is irrelevant as
long as your machine is defined in a DNS server. |
If you are using an auth file with Information Server, the command-line Verity Spider or mkvdk
to index and view documents from Internet Information Server sites which require authorization, and you upgrade to
Information Server V3.61, you must make sure the Basic realm designation is always “host name” and not “IP address.”
Information Server V3.61 sends to Internet Information Server a request header which contains “Host=hostname.” Internet Information Server responds with “WWW-Authenticate: Basic realm = hostname.” Information Server compares the response with the value in the auth file to determine whether or not indexing or viewing can continue. In previous versions of Information Server, the request header could contain an IP address. Now that it is always a host name, you must make sure you use host names in your auth file.
In both Microsoft and Netscape version 4.x browsers, there is a problem with viewing PDF documents when the SEARCHScript which refers to the result document is $$doc.URL_XML.
A workaround is to change the SEARCHScript from:
$$doc.URL_XML
to
$$(doc.URL)a=b#xml=$$doc.XML
The problem is that the browsers are translating the characters “” to the character “?” and this changes
the result URL and does not call the Acrobat Reader to view the PDF document. The SEARCHScript change
preserves the correct URL and will allow you to view the PDF document in your browser.
NOTE: The problem is specific to the browser translation of the characters and has nothing to do with how the PDF files are indexed.
There is a known problem with viewing PDF documents using Information Server 3.61 in NSAPI mode (s97_ns2.so) with Netscape Enterprise Server 2.01 on the IBM AIX unix platform.
As a workaround, use CGI instead. To change from NSAPI to CGI in your search forms and results templates,
change all references to s97is.vts (NSAPI) to search97cgi/s97_cgi (CGI).
Search forms The default search forms available from the Forms Wizard are in
installdir/verity/is/s97is/locale/english/samples/templates
and the default search forms available from the Information Server home page are in
installdir/verity/is/s97is/locale/english/samples/forms
Results templates The results templates used by the default search forms are in
installdir/verity/is/s97is/locale/english/samples/forms
If you installed Information Server V3.61 in NSAPI mode (s97_ns2.so) with Netscape Enterprise Server V2.01 on AIX, there is no support for running Agent Server.
On page 6-2 of the SEARCHScript Reference Guide V3.6, p6-2, the example given for "Viewing Multiple Document Types" is incorrect.
The current string is:
<% if InStr(doc.MIME-Type = "text/html") %>
The corrected string should read:
<% if InStr(doc.MIME-Type, "text/html") %>
Note that there should be a comma, not an equal sign, linking MIME-Type and text/html.