Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-10 Thread Mark Stanton 
It should be an option (though I've not looked at the code), because 
dbf is a *FILE* format that allows for deleted rows to remain in the 
dataset (as deleted).

If you're working with a full database engine you would not (or at 
least, should not) expect that deleted data would remain in an 
accessible state.

Regards
Mark Stanton


___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-07 Thread Alexander Thurgood 
Le 07/04/14 16:24, Lionel Elie Mamane a écrit :

Hi Lionel,

> 
> "used to allow" means that this is not possible anymore?
> 

No, it means that I haven't checked whether it is still available. From
memory, when one had a dbf table, and did a left mouse button click on
the left hand edge of the table view grid control, a context menu was
displayed and one could choose to compact the db, permanently wiping the
deleted tuples, so it wasn't really a toggle on/off UI switch I suppose,
although it looked that way to the user.


Alex

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-07 Thread Lionel Elie Mamane 
On Fri, Apr 04, 2014 at 02:20:41PM +0200, Alex Thurgood wrote:

> In StarOffice 5.x and then OpenOffice.org, the DBF driver used to allow
> a user to see deleted tuples in the data source browser pane and in
> table view mode, and as far as I can remember it was an option in the UI
> that could be switched on/off.

"used to allow" means that this is not possible anymore?

-- 
Lionel
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-05 Thread julien2412 
Lionel Elie Mamane wrote
> It would seem to me that indeed the "ownDeletesAreVisible" is
> wrong. OTOH, it was "==" and it was changed in (by Ocke Janssen, not
> Olivier
> Bolte, who as far as I understand has done the merge):
> ...
> So I would like to be cautious there. I cannot find any definite
> source whether "ownDeletesAreVisible" means "the deleted rows are
> still visible" or "the delete action is visible -> the deleted rows
> are not visible". If we can find some good documentation or example
> from another JDBC driver...

I found a French link (but it's not official),
http://blog.paumard.org/cours/jdbc/chap05-resultset-types.html
"
On peut tester si notre ResultSet voit ses propres suppressions ou pas, par
appel à la méthode DatabaseMetaData.ownDeletesAreVisible(int type). Si elles
le sont, alors la méthode rowDeleted() retourne true si la ligne a été
effacée, false sinon

Il est important de savoir si un ResultSet est sensible ou non aux lignes
que l'on efface dedans, car dans certaines implémentations, les lignes
effacées ne sont pas retirées, mais remplacées par des lignes vides. 
"
quick and dirty translation:
"We can test if our resultset sees its own deletes or not, when calling
DatabaseMetaData.ownDeletesAreVisible(int type). If deletes are visible, so
the RowDeleted() method returns "true" if the row has been removed, false
otherwise. ..

It's important to know if a Resultset takes into account the rows deleted in
it, because in some implementations, the deleted rows are not removed but
replaced by empty rows.
"

Julien



--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005p4104183.html
Sent from the Dev mailing list archive at Nabble.com.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-05 Thread julien2412 
Thank you Lionel for your feedback.
Here's a proposed patch here:
https://gerrit.libreoffice.org/#/c/8863/

Julien



--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005p4104182.html
Sent from the Dev mailing list archive at Nabble.com.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-04 Thread Lionel Elie Mamane 
On Thu, Apr 03, 2014 at 01:28:24PM -0700, julien2412 [via Document Foundation 
Mail Archive] wrote:

> Also, I noticed in this same file:
>1252 sal_Int32 SAL_CALL
> ODatabaseMetaData::getDefaultTransactionIsolation(  ) throw(SQLException,
> RuntimeException, std::exception)
>1253 {
>1254 SQLUINTEGER nValue;
>1255
> OTools::GetInfo(m_pConnection,m_aConnectionHandle,SQL_SUBQUERIES,nValue,*this);
>1256 return nValue;
>1257 }

> Searching about getDefaultTransactionIsolation, I found
> SQL_DEFAULT_TXN_ISOLATION. But this attribute isn't in sqlext.h but in
> isqlext.h (see
> http://sourcecodebrowser.com/unixodbc/2.2.11/nn_2isqlext_8h.html)
> Nevertheless, looking the number of locations where sqlext.h is used in LO,
> I wonder what impact would be including isqlext.h
> Still, the above code with SQL_SUBQUERIES seems weird.

Why would there be any problem in including isqlext.h? Just do it...

Yes, the above call with SQL_SUBQUERIES looks obviously wrong, and
needs to be replaced by SQL_DEFAULT_TXN_ISOLATION + translating from
ODBC constants (SQL_TXN_READ_UNCOMMITTED, SQL_TXN_READ_COMMITTED,
SQL_TXN_REPEATABLE_READ, SQL_TXN_SERIALIZABLE) to SDBC constants
(com::sun::star::sdbc::TransactionIsolation).

-- 
Lionel




--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005p4104094.html
Sent from the Dev mailing list archive at Nabble.com.___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-04 Thread Alex Thurgood 
Le 04/04/2014 13:13, Lionel Elie Mamane a écrit :

Hi,


> So I would like to be cautious there. I cannot find any definite
> source whether "ownDeletesAreVisible" means "the deleted rows are
> still visible" or "the delete action is visible -> the deleted rows
> are not visible". If we can find some good documentation or example
> from another JDBC driver...

In StarOffice 5.x and then OpenOffice.org, the DBF driver used to allow
a user to see deleted tuples in the data source browser pane and in
table view mode, and as far as I can remember it was an option in the UI
that could be switched on/off.


Alex


___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-04 Thread Lionel Elie Mamane 
On Thu, Apr 03, 2014 at 02:02:08PM -0700, julien2412 [via Document Foundation 
Mail Archive] wrote:

> I found another part however:
>1622 sal_Bool SAL_CALL ODatabaseMetaData::ownDeletesAreVisible( sal_Int32 
> setType ) throw(SQLException, RuntimeException, std::exception)
> ...
>1641 return (nValue & SQL_CA2_SENSITIVITY_DELETIONS) != 
> SQL_CA2_SENSITIVITY_DELETIONS;
>1642 }

> whereas we have:
>1644 sal_Bool SAL_CALL ODatabaseMetaData::ownInsertsAreVisible( sal_Int32 
> setType ) throw(SQLException, RuntimeException, std::exception)
> ...
>1663 return (nValue & SQL_CA2_SENSITIVITY_ADDITIONS) ==  
> SQL_CA2_SENSITIVITY_ADDITIONS;
>1664 }

> (See
> http://opengrok.libreoffice.org/xref/core/connectivity/source/drivers/odbc/ODatabaseMetaData.cxx#1622)
> One of them must be wrong.


It would seem to me that indeed the "ownDeletesAreVisible" is
wrong. OTOH, it was "==" and it was changed in (by Ocke Janssen, not Olivier
Bolte, who as far as I understand has done the merge):


commit bb075c9e3f2f8e0e9ce073a3dad5201f4e5fcc91
Author: Oliver Bolte 
Date:   Fri Mar 18 08:57:23 2005 +

INTEGRATION: CWS dba25 (1.27.204); FILE MERGED
2005/03/04 11:37:35 oj 1.27.204.2: simple correct for ownDeletesAreVisible
2005/03/04 10:12:19 oj 1.27.204.1: correct operator order


So I would like to be cautious there. I cannot find any definite
source whether "ownDeletesAreVisible" means "the deleted rows are
still visible" or "the delete action is visible -> the deleted rows
are not visible". If we can find some good documentation or example
from another JDBC driver...

-- 
Lionel




--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005p4104066.html
Sent from the Dev mailing list archive at Nabble.com.___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

Re: About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-03 Thread julien2412 
Re reading part 1), i was obviously wrong.
SQL_ORDER_BY_COLUMNS_IN_SELECT indicates if columns contained in order by
clause must be in Select part.

So it's normal supportsOrderByUnrelated must test this value equal to 'N'


I found another part however:
   1622 sal_Bool SAL_CALL ODatabaseMetaData::ownDeletesAreVisible( sal_Int32
setType ) throw(SQLException, RuntimeException, std::exception)
...
   1641 return (nValue & SQL_CA2_SENSITIVITY_DELETIONS) !=
SQL_CA2_SENSITIVITY_DELETIONS;
   1642 }

whereas we have:
   1644 sal_Bool SAL_CALL ODatabaseMetaData::ownInsertsAreVisible( sal_Int32
setType ) throw(SQLException, RuntimeException, std::exception)
...
   1663 return (nValue & SQL_CA2_SENSITIVITY_ADDITIONS) ==
SQL_CA2_SENSITIVITY_ADDITIONS;
   1664 }

(See
http://opengrok.libreoffice.org/xref/core/connectivity/source/drivers/odbc/ODatabaseMetaData.cxx#1622)
One of them must be wrong.

Julien



--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005p4104014.html
Sent from the Dev mailing list archive at Nabble.com.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice

About parts in ODatabaseMetaData.cxx (connectivy module/odbc part)

2014-04-03 Thread julien2412 
Hello,

Taking a look at connectivity/source/drivers/odbc/ODatabaseMetaData.cxx
without particular reason, I noticed this:
   1044 sal_Bool SAL_CALL ODatabaseMetaData::supportsOrderByUnrelated(  )
throw(SQLException, RuntimeException, std::exception)
   1045 {
   1046 OUString aValue;
   1047
OTools::GetInfo(m_pConnection,m_aConnectionHandle,SQL_ORDER_BY_COLUMNS_IN_SELECT,aValue,*this,m_pConnection->getTextEncoding());
   1048 return aValue.toChar() == 'N';
   1049 }
See
http://opengrok.libreoffice.org/xref/core/connectivity/source/drivers/odbc/ODatabaseMetaData.cxx#1044

Whereas all the other blocks compare to 'Y', I don't understand why it
should be compared to 'N' and not to 'Y' like the other blocks.


Also, I noticed in this same file:
   1252 sal_Int32 SAL_CALL
ODatabaseMetaData::getDefaultTransactionIsolation(  ) throw(SQLException,
RuntimeException, std::exception)
   1253 {
   1254 SQLUINTEGER nValue;
   1255
OTools::GetInfo(m_pConnection,m_aConnectionHandle,SQL_SUBQUERIES,nValue,*this);
   1256 return nValue;
   1257 }

Searching about getDefaultTransactionIsolation, I found
SQL_DEFAULT_TXN_ISOLATION. But this attribute isn't in sqlext.h but in
isqlext.h (see
http://sourcecodebrowser.com/unixodbc/2.2.11/nn_2isqlext_8h.html)
Nevertheless, looking the number of locations where sqlext.h is used in LO,
I wonder what impact would be including isqlext.h
Still, the above code with SQL_SUBQUERIES seems weird.

Any idea?

Julien



--
View this message in context: 
http://nabble.documentfoundation.org/About-parts-in-ODatabaseMetaData-cxx-connectivy-module-odbc-part-tp4104005.html
Sent from the Dev mailing list archive at Nabble.com.
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice