On 10 Nov 2002 at 23:11, Geoff Hutchison wrote: > > On Saturday, November 9, 2002, at 04:27 PM, Dan Langille wrote: > > > years of data. I'm seeking comments on my approach. > > You're doing some unnecessary moving around, but I'd need some > clarifications. I *think* what you're doing is indexing into the -merge > databases, which are *only* used for indexing and the eventual merging > into the old database. And the main database is only used for searching > and the merging. > > Right?
Correct. > > cp adsl-merge.docdb.work adsl-merge.docdb > > cp adsl-merge.docs.index.work adsl-merge.docs.index > > cp adsl-merge.wordlist.work adsl-merge.wordlist > > cp adsl-merge.words.db.work adsl-merge.words.db > > OK, if you're only ever using this for merging, then this is completely > useless. Only the .work files are ever touched. So you have a lot of > duplicate data. I tried the procedure without this step. It failed to include the new documents to the final index. When I tried again, doing the cp, it found the new document. I don't yet have an explanation for this but I'm calling it a day for now. I'll look into it later. FWIW, the full script is at http://www.unixathome.org/index-merge-rundig-test.sh.txt > > After the merge, this moves the new search data into production: > > mv adsl.docdb.work adsl.docdb > > mv adsl.docs.index.work adsl.docs.index > > mv adsl.wordlist.work adsl.wordlist > > mv adsl.words.db.work adsl.words.db > > OK, but the .wordlist file is never used by htsearch. So you might as > well leave it as a .work file (where it's used by the merge) and never > copy it either. > > Also, if you have the disk space, you can change those "mv" commands to > "cp" and leave the .work files in place--it'll save time, though > admittedly at the expense of disk space. Good idea. But I will do a mv, then cp them back. This will reduce the chance that htsearch is invoked on a file which is being copied to. -- Dan Langille : http://www.langille.org/ ------------------------------------------------------- This sf.net email is sponsored by:ThinkGeek Welcome to geek heaven. http://thinkgeek.com/sf _______________________________________________ htdig-general mailing list <[EMAIL PROTECTED]> To unsubscribe, send a message to <[EMAIL PROTECTED]> with a subject of unsubscribe FAQ: http://htdig.sourceforge.net/FAQ.html
