The next upgrade to Koha will be a major re-write of Koha's indexing platform. The new software is now available for testing in the SCLS Koha sandbox (both staff client and PAC). If you would like to try it out, you can log into the sandbox using the same login and password that you use for Koha production.
The sandbox is a work in progress and additional fixes are still being added. After we have an upgrade date, we will post full release notes on the ILS web site.
Why are we switching to Solr?
Solr is an open-source search indexing platform that is faster and offers more flexibility than Koha's current indexing platform (Zebra). It fixes a number of searching bugs and will offer us the ability to create browse indexes in the future.
What are the new features and bug fixes that are coming with Solr?
Here is a list of fixed bugs and new features:
- Indexing of new bibs and items: With Solr, new bib records and item records will be indexed in approximately 5 minutes, instead of several hours. Edits to existing bib records and items will also index in approximately 5 minutes. (SCLS Support Ticket 39738)
- Improvements to relevancy ranking: Exact main title (245$a) matches are ranked before a main title with the same words, but that is not an exact match. Also, English-language books continue to be ranked above AV and foreign language materials. For example, a title search for The passage will return titles with that exact main title ranked at the top of the list. These titles will be ranked about titles such as Spirit of the passage and The passage of power, which have the same words but are not exact matches. Please note that exact matches are prioritized, so a search for The passage will rank a record titled The passage before a record titled The passage : a novel. In some cases this will cause an AV record to be ranked above a book record. For example, if a record for a book contains a sub-title but the AV record does not have that sub-title, then the AV record will be ranked above the book record, because Koha will rank the main title first, and then rank the English-language books. Finally, these relevancy ranking improvements have been applied to both the title and keyword indexes, as well as the Cataloging search in the staff client.
- Problems with Author search: The author searches in both the staff client and PAC are now working correctly. Previously, the Author searches in the drop-down box on the main page of the PAC and in the advanced search of the staff client were searching title fields as well as author fields. (SCLS Support Ticket 41891)
- Facets work for entire search results list: Facets now include results from the records in the entire search results list. Previously, the facets were generated based on only the current page of search results. (SCLS Support Ticket 40742)
- Searching of titles and authors with diacritics: You can now search titles and authors even if there is a diacritic in the title or author fields. Previously, these searches were failing. (SCLS Support Ticket 40822)
- Using ampersands in searches: Using an ampersand in the search no longer causes problems with browsing search results. Previously, if you used an ampersand in a search, then you lost the search results when you went to the second page. (SCLS Support Ticket 42602)
- Keyword searching in staff client for titles that are numbers: Using quotes allows you to do a keyword search in the staff client for titles that are numbers. For example, search “1493” to find the title 1493 : uncovering the new world Columbus created. (SCLS Support Ticket 41768)
- Re-sorting the search results list in the staff client: Previously, re-sorting the search results list did not always work. Instead, you were taken to the Advanced Search page. This only happened when you used an equal sign in the search (e.g. au=john grisham). Because you now must use a colon when you are using a search operator, this problem is no longer an issue (e.g. au:john grisham). (SCLS Support Ticket 39838)
- Punctuation in titles: Searches that include titles with punctuation, such as colons, slashes, exclamation points, etc. no longer fail. Sometimes you may need to use quotes around the title, but you should not need to strip punctuation.
Are there any new search tips?
Yes, Solr offers more tools for refining searches:
- Use double quotes to search an exact phrase in any index. (Single quotation marks are ignored). For example, searching "lord of the rings" will only return records that contain that exact phrase. (NOTE: We will no longer have separate phrase indexes for title, author and subject. These indexes have been replaced by the double quotes feature).
- Place your search terms in parenthesis to return search results on words in any order. For example, searching (lord of the rings) will return records that contain all of those words, but not necessarily in that order.
- Use a question mark for single character right-hand truncation, for example to find nap or naps, search nap?.
- Use a star for multiple character right-hand truncation, for example to find nap or naps or napping, search nap*.
- Use a question mark or star for center truncation, for example to find woman or women, search wom?n [single character] or wo*n [multiple character].
- Use a tilde combined with double quotes to make two terms appear within a certain number of characters of each other, for example "roman empire"~10 would search for the two words within 10 characters of each other.
- You can use the following search operators when using the keyword index to narrow your search. You do not have to use the search operators when using the keyword index – they are tools that are available but they are not mandatory. You must use a colon (:) between the search operator and the search terms, (e.g. ti:the help):
- ti: (title)
- au: (author)
- su: (subject)
- isbn: (isbn)
- issn: (issn)
- biblionumber: (biblionumber)
Are there any known bugs in the sandbox that aren't in production?
Yes, there are some bugs in the sandbox that are not present in production. These bugs will be fixed before we upgrade:
- Library limit (Location and Available limit) in staff client doesn't work. Ok in PAC.
- When searching ISBN-13, ISBN-10 is not found (and vice versa).
- Clicking "more" link in facets column generates error in staff client. Ok in PAC.
There may be more bugs present in the sandbox (that are not already in production). See the next section for information on helping out with testing.
How can I help with testing Solr?
Fill out the Solr testing checklist. If you cannot finish the entire checklist, that's fine, just do what you can and leave the other questions blank. Please complete any testing by Monday, October 1.
You can also report problems that you encounter in the sandbox using the Koha problems and questions form.