Page 3 of 12
Posted: 27.05.2013, 16:46
by EgonHugeist
miab3,
i got the little bug! Patch done R2270.
IMHO was it an all time issue. I made a SVN compare and the reported issue was never handled before. Can you confirm my patch?
I didn't touch the array types. Still fighting with the updates for minior String-convertations...
Posted: 27.05.2013, 17:51
by miab3
@EgonHugeist,
After R2270 it's fine.
Michal
Posted: 04.06.2013, 20:50
by miab3
@EgonHugeist,
Could you change in ZDbcDbLibMetadata.pas:
Code: Select all
+ ' sysobjects o on (o.id = c.id) where o.name = %s and c.number=0 order by colid',
on
Code: Select all
+ ' sysobjects o on (o.id = c.id) where o.name COLLATE Latin1_General_CS_AS = %s and c.number=0 order by colid',
I will not have to constantly fix it.
This is an explanation:
http://blog.sqlauthority.com/2007/04/30 ... ry-search/
Michal
Posted: 08.06.2013, 21:15
by EgonHugeist
miab3,
Patch done R2289 \testing7.1 (SVN)
Thank you Michal.
May i ask you if you have some benchmark tests to check a internal speed increase?
Posted: 09.06.2013, 12:29
by miab3
@EgonHugeist,
I did not notice a significant difference in the speed of the R2290.
[s]However, recently I noticed a drop in performance ORACLE particular read/update master-detail-subdetail (and i not are able to determine which patch caused it).
(A similar test at Firebird is three times faster than ORACLE).[/s]
I think it the fault of my ORACLE (I'll have to put it all over again).
Michal
Posted: 09.06.2013, 14:04
by EgonHugeist
miab3,
that's not what i mean. I'm currently working on the RawUpgrade and do optimize the String convertion routines. Tody i've killed my local working copy of the dbc-layer. Inbetween i'm a little bit frustrated. Well i'll try to merge "save" changes to SVN as good as possible (since i've again over 3000 changes lines and fixing the current code becames a merging hell).
But i could also send you a copy before commit the changes to verify if it works and even if it's faster. Currently i can't say when i'm ready. Mainproblems i have are CodePage inconsistend CodePages like FireBird CharacterSet "NONE". Where each column can have a different CodePage. Currently i've no idea how to tell the RowAccessor which col uses which CP. (not a problem on DBC-Layer but a huge problem on the Component layer).
So i think it could take a while (one or two weeks). And even more would it be interesing if this patch speeds up something. ??
Posted: 09.06.2013, 18:57
by miab3
@EgonHugeist, @mdaems
It might be worthwhile to publish the current version 7.10 as stable before you start a revolution.
(It looks me the stable, more stable than 7.03).
Michal
Posted: 11.06.2013, 11:29
by miab3
@mdaems, @EgonHugeist
Zeos7.1-alpha branches_ testing_R2300:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
Michal
Posted: 16.06.2013, 00:10
by EgonHugeist
miab3,
It might be worthwhile to publish the current version 7.10 as stable before you start a revolution.
(It looks me the stable, more stable than 7.03).
I found and fixed a loads of remaining issues on 7.0/7.1. But you're right. We should relase 7.0.4-stable and go beta with 7.1. Mark and me already wrote about..
Posted: 16.06.2013, 12:03
by miab3
@mdaems, @EgonHugeist
Zeos7.1-alpha branches_ testing_R2347:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
Two little remarks(from D2006):
[Pascal Hint] ZVariant.pas(1939): H2443 Inline function 'ZConvertStringToRawWithAutoEncode' has not been expanded because unit 'WideStrUtils' is not specified in USES list
[Pascal Warning] ZVariant.pas(1926): W1035 Return value of function 'TZClientVariantManager.GetAsRawByteString' might be undefined
Michal
Posted: 16.06.2013, 20:26
by miab3
@mdaems, @EgonHugeist
Zeos7.1-alpha branches_ testing_R2358:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
Michal
Posted: 18.06.2013, 11:41
by miab3
@mdaems, @EgonHugeist
Zeos7.1-alpha branches_ testing_R2374:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
This fix(R2374) improves in the D2006/D2007 and in XE2 too.
Michal
Posted: 18.06.2013, 12:25
by EgonHugeist
miab3,
thanks, Michal. Yes this was a nasty bug. Equal to R2375 (;
Now we could also assume ftCurrency for fields with equal or less decimal digits of 4. What do you think?
Btw.
http://zeosdownload.firmos.at/downloads/snapshots/ its just a snapshot. Mark will merge the latest fixes and 7.0.4 (a bugfix release) will be released the next days...
7.1 goes beta next days too (:
Posted: 18.06.2013, 22:28
by miab3
@EgonHugeist, @mdaems,
Zeos7.1-alpha branches_ testing_R2375:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
Michal
There is a NEW comparison ZEOS-FireDAC-UniDAC:
http://blog.synopse.info/post/2013/06/1 ... for-mORMot
Posted: 20.06.2013, 15:49
by miab3
@EgonHugeist, @mdaems,
Zeos7.1-alpha branches_ testing_R2381:
http://svn.code.sf.net/p/zeoslib/code-0 ... sting-7.1/
compiles and runs on (I have tested):
- D2006,
- D2007 -> C++,
- DXE2 32/64 -> C++(32),
- Lazarus Windows7-64 1.1.0/fpc 2.7.1 32/64 (CodeTyphon ver 4.30),
(For Linux currently I have no way to test)
(Firebird 2.5.3, MySQL 5.5.18, MariaDB 5.5.29, PostgreSQL 9.2.2, MSSQL 2008 R2 Express(mssql, FreeTDS, ado-sqlncli), *.mdb-ado-Jet.OLEDB.4.0, Oracle Database Express Edition 11g Release 2, SQLite 3.7.17).
Michal