1 |
On 2012-09-17 10:25 AM, Michael Mol <mikemol@×××××.com> wrote: |
2 |
> On Mon, Sep 17, 2012 at 9:46 AM, Tanstaafl<tanstaafl@×××××××××××.org> wrote: |
3 |
>> Never mind, just got a reply from the dev that he had fixed it and the next |
4 |
>> update would contain the fix... |
5 |
>> |
6 |
>> I'm still curious if I should not go down that road and use something else |
7 |
>> for FTS... |
8 |
|
9 |
> clucene looks like the thing to use right now, unless you want to use |
10 |
> the full Java-based Apache Lucene. I was just looking at it this |
11 |
> morning as a possible basis for a solution to a problem of my own[1], |
12 |
> since strigi uses it. |
13 |
|
14 |
Thanks Michael... |
15 |
|
16 |
Just to wrap up this thread, in case anyone is interested, because Timo |
17 |
(dovecot author) had replied in a similar thread on the dc list that he |
18 |
thought there was talk of merging clucene and lucene++, I queried the |
19 |
clucene author after he replied he had fixed this issue, and here is his |
20 |
reply: |
21 |
|
22 |
"More or less it's true. About a year ago we started to make Lucene++ to |
23 |
the new CLucene version, as Lucene++ (also written in C++) is a port of |
24 |
a newer Apache Lucene version (written in Java) as the one CLucene is a |
25 |
port of. But we did not want to simply merge them, but to adapt Lucene++ |
26 |
to the "design principles" of CLucene. E.g., Lucene++ makes heavy use of |
27 |
shared pointers. And in CLucene we wanted to reduce this usage in favor |
28 |
of performance. But this not finished and I cannot say when it will |
29 |
finished. Nevertheless, the new version of CLucene (if any) will be also |
30 |
C++ and not Java. Best regards, Veit" |