Re: Search fails, but only for a specific sequence

Tech-Archive recommends: Fix windows errors by optimizing your registry

On Feb 14, 9:52 am, Puddles <Pudd...@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
Hallo Ginny,

like I said, a SPECIFIC production of my database displays this unusual
feature: either you take this database in it's completness or nothing! A
connection from SQL Server Managemant Studio is all you need to issue a
select with the specific LIKE values.

"Ginny Caughey [MVP]" wrote:
What I need is for you to provide a very small app that creates the database
and adds enough values to demonstrate the issue. I'm not doubting your word
or anything like that, I just need something to pass along to the
developers. I'd be interested in any characters in the key column with
values > 127, but whatever it takes to reproduce the problem is what I need.

Thanks,

--
Ginny Caughey
Device Application Development MVP

www.wasteworks.com
Scalehouse and Billing Software for Waste Management

"Puddles" <Pudd...@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:FF985CFA-AE29-4AC7-BAAA-07FB5F6B28B3@xxxxxxxxxxxxxxxx
Hallo Ginny,

Obviously I could give you my database to test - but it is over 50Mb big!
Please let me know how I can deliver this.

I did look at the data relevant to an index over this area of the data and
I
can assure you there was nothing fishy there! Basically, if the data was
bad, why is it being delivered without a problem using the shorter LIKE
string? Apart from that, what iconstitutes a special character in a
Unicode
database?????

"Ginny Caughey [MVP]" wrote:

Well, TableDirect is a bit of a secret ;-) but I think people who do know
about it don't use if often (including me) because using T-SQL is more
natural for most developers and easier. It is fast though, as you found,
and
in low memory situations it can be the only viable option.

I don't know if your index was faulty or not since you didn't show a list
of
the data values that were indexing "incorrectly". If the index is wrong,
I
can report that to Microsoft to get it resolved, but they'll want to see
a
reproducible example.

--
Ginny Caughey
Device Application Development MVP

www.wasteworks.com
Scalehouse and Billing Software for Waste Management

"Puddles" <Pudd...@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:F91267BD-9F67-471D-96E6-58FE5BEE94F2@xxxxxxxxxxxxxxxx
Hallo Ginny,

neat tool, this TableDirect... how come I had never discovered it in
over
three years of Mobile development? Huge speed advantage over SQL
queries,
but (and it is a big 'but') there is no easy way to implement DISTINCT.
Three options are 1. a new table with only unique elements (blows the
database size), 2. filter row for row in C# (slow) or 3. use the
ToTable()
function of a DataView (memory hungry). Hummm, now it looks like the
SQL
query is infact the better option!

All very interesting and my original issue has solved itself anyway: a
new
production with changed data (happens every 14 days) - in effect the
same
as
producing the index using DESC. But I think this is ignoring the real
problem - the index was faulty and this must be regarded as a bug by
the
developers at Microsoft! Who know where it will appear next!

"Ginny Caughey [MVP]" wrote:

Puddles,

You wouldn't load the entire table into memory with TableDirect since
you'd
specify an index name and range. But in any case since adding DESC
changes
the behavior to the one you want, I'd expect there to be a value
somewhere
between 'ne' and 'ney' that is causing the result you see. I think if
you
look at the names in that range you'll get more insight what exactly
is
happening. If you do, be sure to tell the rest of us!

--
Ginny Caughey
Device Application Development MVP

www.wasteworks.com
Software for Solid Waste Management

"Puddles" <Pudd...@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:46551D72-D5A7-4039-89E8-AA78803205B1@xxxxxxxxxxxxxxxx
Hallo Ginny,

Using TableDirect would imply moving the entire table into the PDA's
memory
and paring it there with C#... then why bother with the database?

I do not think there are any strange characters causing the system
to
go
crazy - the looser search for 'ne%' works and includes everything
that
should
be returned by 'ney%'!

I have created a workaround by adding the DESC parameter to the
index
definiton (yes, the name colunm has it's own index), though I rather
fear
this has simply moved the bug to a new location (I wonder which
customer
will
discover it first?).

"Ginny Caughey [MVP]" wrote:

What are the medicines you would expect to find in that range of
values?
I'm
guessing that accented characters specific to German might be
causing
the
result you're seeing.

Assuming that you have an index on Name, do you get the same result
using
TableDirect access? The reason I ask is that using TableDirect no
longer
uses the query processor but it does still use the index order.

--
Ginny Caughey
Device Application Development MVP

www.wasteworks.com
Software for Solid Waste Management

"Puddles" <Pudd...@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:2295EF7D-6EF7-480F-A838-AFB11C0F2DFF@xxxxxxxxxxxxxxxx
This is weird! My medical database works just fine and has done
since
ages,
until someone treid to look for a medicin begining with 'ney....'

Major Error 0x8007007A, Minor Error 0
SELECT * FROM t_artikel WHERE NAME LIKE 'ney%'
Der an einen Systemaufruf übergebene Datenbereich ist zu klein.

OK, so it's a German installation, but the issue is not hardware
relevant:
same on any Mobile device and on my PC. Varying the search
string
usually
works, like 'ne%' is OK and 'n_y%' is ok, but 'new%' fails as
does
'ney_%'.

Can anyone tell me what is going on here?

Thanks in advance.

If it's a specific DB among many that behaves this way, it could be
due to corruption.
DB's have been known to get corrupted, especially the index.

- Jin
.

Relevant Pages

  • Re: Search fails, but only for a specific sequence
    ... a SPECIFIC production of my database displays this unusual ... Device Application Development MVP ... Scalehouse and Billing Software for Waste Management ... You wouldn't load the entire table into memory with TableDirect since ...
    (microsoft.public.sqlserver.ce)
  • Re: Is it just me or are there BIG problems with SQLCE 3.0?
    ... It would also be important to have a repro example to test against 3.5 as well. ... Device Application Development MVP ... Hitchhiker's Guide to Visual Studio and SQL Server ... use the Open update Close technique, the database is compressed at the final ...
    (microsoft.public.sqlserver.ce)
  • Re: How to read data from stand alone SQL CE database on mobile device from VB.net 2005
    ... Device Application Development MVP ... consider removing the IIS requirement for SQL Server but I don't ... I have a stand alone SQL CE database on my windows mobile device. ...
    (microsoft.public.sqlserver.ce)
  • Re: Multiple User Access through ADO?
    ... SQL Compact also can act as a Subscriber in a Merge replication scenario or synchronize with a common database using the Local Data Cache/ADO.NET Sync Services. ... "Ginny Caughey MVP" wrote in message ... Device Application Development MVP ...
    (microsoft.public.sqlserver.ce)
  • Re: SDF Database
    ... I'm using CTP6 still, and I can alter the schema ok, but I know of at least one person who had problems with a later version that sounds similar to what you're seeing. ... Device Application Development MVP ... I've checked the database with Chris' code, it is V 3.5 before and after opening in VS 2008... ... Ok I can open it now in SQL Management Studio 2008, but have to use sql statements there to alter it. ...
    (microsoft.public.dotnet.framework.compactframework)