[sword-devel] Windows beta 1.5.8pre3

Lynn Allan l_d_allan at adelphia.net
Sat Feb 12 05:42:14 MST 2005


> I did hit an access violation specifying the 2nd Bible for Parallel:
> "Access violation at address 0040913D in module
'sword-1.5.8pre1.exe'.
> Read of address 00000000."

These seem fixed with pre3. Thanks.

> I tried the "Create Index" for searching the ACV. It is still
grinding
> and taking a loooooong time ... almost seems locked up, but I'll
wait
> a while longer. Every minute or so, another block appears in the
> "Progress Indicator" so this could take a while.  Seems a little
more
> than half done after 13 minutes on an older 933 mhz PIII ... done @
25
> minutes. Is this expected behavior?

pre3 seems slower to build an index ... now up to 32 minutes for ACV.

> Hmmmm ... I don't see how to specify I want all three words of "Lord
> Jesus Christ" to appear. The radio buttons for "Multi Word",
"Phrase",
> and "Regular Expression" are greyed. Indexing seems to disable these
> radio buttons ... they are available with a Bible text that hasn't
> been indexed (see AKJV below)

The radio buttons are enabled now after an index has been built. I see
a new option for "optimized search". I'm not clear what these do, but
will investigate.

With ISV, "lord jesus christ" and "jesus lord christ" now indicate 88
matches using "MultiWord" radio button. This behavior seems closer to
expected with an implied "AND". However, no verses show up in the
bottom "widget". Is the 1.5.6 "widget" for "Reference + Preview" gone?

> Indexing should be slightly faster.  Found some issues with CLucene
> still running out of file handles.  I think it's fixed and this has
made
> lucene build an index faster. Unfortunately, you'll have to delete
your
> current indecies PLEASE DON'T TRY USING THIS WITH INDECIES BUILD
WITH pre1.


Don't know if this helps ... with pre3, I used Win2000-Sp4's
TaskManager to watch the "handle" count while the ACV index was being
deleted and then rebuilt.

I'm guessing that "handle count" refers to files opened, but I could
be mistaken. Anyway, there weren't that many handles in use (55-80),
but there seemed like a lot a activity. At times, the count would
change from 55 to 80 and back every second or so. Then it would seem
to "settle down" to the range of 55-57.

I'll watch another index build from the start and provide info. Don't
know if this helps diagnose anything.

Is there anything interesting/useful to watch with "Perf" that might
be helpful?

On my system, the IndexBuild of ACV seems to be slower on pre3 than
pre1. It has more or less locked up my system. I can still type into
this email using Outlook Express 6, but pre3 and other apps are "Not
Responding". I can't get pre3's Main window for Search Dialog to come
to the foreground to see how far it has progressed

Getting ready to BuildIndex for ISV. TaskMgr shows 81 handles in use
before indexing. Dropped to 55. Progress bar immedediately shows about
3/4 done, but then near-lockup of entire computer (as described
above). Can't 'drag' titlebar of Search around. Clicking on main Sword
window's TitleBar causes "Search" titlebar to be selected, but it
doesn't come to the foreground and isn't responding. Handle count is
"jumping around" every second from 55-60-80-55-56-79 etc on an
apparently random basis. Using about 11 meg of memory, which doesn't
seem excessive.

I "turned on a lot" of counters in "TaskManager". At about 7 minutes
into ISV BuildIndex:
I/O Reads: 740,000
I/O Writes: 36,000
I/O Other: 2,200,000
I/O ReadBytes: 75,000,000

Does this mean anything? About what you expected, or higher?

> Took another 25 minutes to finish indexing ISV ... but it did
finally
> get done.
>
> I would have thought it would take less time since the ISV is only
the
> NT and the ACV has both the OT & NT.
>
ISV got done after about 8 minutes, which is MUCH faster than before
(25 minutes on pre1 ... 933 mhz P3 with 768 meg PC100 memory).

> Odd ... before indexing, Search found 1180 instances of "Jesus" in
the
> ISV  .... finds 1179 instances after indexing???

Searching for "Jesus" before and after index: 1180 matches. Fixed.

There is a LOT (hundreds? thousands?) of redrawing of the "slider"
near the bottom (under the "Create VerseList From Result". Nothing
shows up in the bottom "widget". What is it for?

Odd ... after the Index, the selected text changed from ISV to
something else (KJV Strongs?). I had trouble reselecting ISV using the
"Choose Module" widget ... had to close the Search dialog and reclick
on Search to search ISV.

Perhaps odd ... I started a search for "jesus lord christ" as "Phrase"
using KJV+Strongs unindexed. Taking a long time (about 40 seconds), as
expected because KJV-Strongs not yet indexed.

Mid way through out of curiousity and bored impatience, I clicked on
ISV on the left "Choose Module" TreeControl. The Search kept grinding
and the progress bar advanced. At the end, the TitleBar showed 0
matches from ISV. Odd. Bug?

Should the TreeControl for "Choose Module" be diabled? The "info" next
to the "Create Index" shows "Not using Indexed Searching Framework for
This Book", even though ISV has indices. (KJV-Strong's doesn't yet).
Seems like some kind of "refresh" is left out.

I then click on Search, and it quickly finds no matches for
KJV-Strongs, which is probably confused. I think it found no matches
for ISV, but reported for KJV-Strongs because a KJV-Strongs unindexed
search would be slow.

Search reports 4 matches for "hath" but it's a bit difficult to tell
what they are. I used "Create Verse List" and see 01322, 02188, 04558,
and 05220 with no text. I guess these are OT since they are relatively
low verse numbers?.

When I double click on a number, it takes me to the main window and
attempts to show these verses for the ISV, which of course does not
have the OT. In any case, I was expecting hundreds and hundreds of
matches for KJV-Strongs "hath" and (perhaps?) none from ISV, depending
on whether "exact match entire word" was in use.

Verse on new line seems to be working fine. Thanks.

Paragraphs? (sorry to ask)

HTH and thanks for all your efforts on this.




More information about the sword-devel mailing list