[sword-devel] 1 bug left for 1.5.8
dmsmith555 at gmail.com
Thu Mar 31 17:26:38 MST 2005
For the indexing do you want to index all words (SimpleAnalyzer) or stop
on common words (as defined by lucene) (i.e. StandardAnalyzer). For
JSword, we settled on SimpleAnalyzer. Originally we did the
StandardAnalyzer, but felt that we should control the stop words
external to the index. It took the NASB index from 2M to 2.6M. It seemed
to me that the index times were roughly the same, though Simple seemed a
bit faster and less resource intensive than Standard.
Also, when building the index it is possible to throttle back the number
of open files. This may improve performance on smaller machines.
Neither of these are bugs, but should be considered before release.
Daniel Glassey wrote:
>> ATM morph entry attributes are only parsed for certain markup types,
>> not for all. Please make this work consistently, as BibleTime makes
>> use of the EntryAttributes Search, and it would be nice if the next
>> version had full support for it. Thanks!
>> All Comments Change History Comment by Troy A Griffitts
>> [20/Mar/05 01:15 PM] [ Permlink ] Martin, Could you comment further
>> on this issue. I thought we fixed this in Cambridge. Let me know
>> specifically what needs fixed.
>> Comment by Martin Gruner [23/Mar/05 04:07 AM] [ Permlink ] well, it
>> id definitely not fixed yet. Morph entry attribute parsing seems to
>> work partly in OSIS:
>> lookup KJV "Gen 1:1"
>> looks correct. But
>> lookup KJV "John 1:1"
>> shows that the OSIS filters don`t parse morph stuff in the NT.
>> lookup RST "John 1:1" (similar with LXX)
>> shows that GBF filters don`t parse it at all, and the stron`s entry
>> attribute are partly messed up (contain the morph codes as well).
>> lookup TR "John 1:1"
>> shows that in ThML the parsing is incorrect (afaict, the text
>> Thus EntryAttribute parsing is basically broken for morph codes atm.
>> If this is to be a public interface of the API, it should be fixed.
>> I would personally appreciate it a lot, I think searching for Morph
>> codes is a neat feature that enables lay persons to interact with the
>> grammar of the original documents without knowing the original
>> language itself. Dunno how much work fixing this would be, perhaps it
>> is just a few lines that contain mistakes?
> This is the details of the only bug left in the library for 1.5.8. If
> this bug isn't important enough please move it to a future release. If
> anyone can fix it please do. If there are any library bugs that aren't
> in the tracker please put them there or 1.5.8 will be released without
> a fix.
> sword-devel mailing list: sword-devel at crosswire.org
> Instructions to unsubscribe/change your settings at above page
More information about the sword-devel