Activity for Sergei Kuchin
-
Sergei Kuchin posted a comment on discussion OTL
Jose, You can email me directly. My email address can be found on OTL Web pages. Cheers, Sergei
-
I put together the following quick test code to test timestamp: include <iostream></iostream> using namespace std; define OTL_ORA18C include "otlv4.h" int main(int argc,char* argv[]) { try{ otl_connect db; db.rlogon("system/oracle@myora_tns"); db.direct_exec("drop table test_tab",0); db<<"create table test_tab(update_flag varchar2(1), ipodate timestamp(6), f1 CLOB, f2 CLOB) "; otl_stream o; o.open(1, "INSERT INTO test_tab VALUES( " " :UPDATEFLAG_<char<span>[3]>, " " to_timestamp(:IPODATE<char<span>[27]>,'yyyy-mm-dd...
-
Jose, Can you put together a self-contained code example that reproduces the problem and email it to me? I need a code example that creates a similar table and tries to insert a row into it. BTW, did you enable #define OTL_ORA_TIMESTAMP: https://otl.sourceforge.net/otl3_compile.htm#OTL_ORA_TIMESTAMP Sergei
-
Jose, I'm not sure where / what the problem is. You can always use a char[XXX] variable bind variable + a conversion function call to convert strings into timestamps for passing in timestamp values if nothing else works. It may take weeks to track down the actual issue, and it may take hours to implement a char[XXX] workaround. Sergei
-
Natalia, I looked at that a long time ago but I found that there wasn't enough demand for it at that point in time. At this point in time I don't have "the bandwidth" (too much stuff to do at work and C++ compiler development being "agile" these days doesn't help) to implement anything as complex as bulk insert. I'd suggest you to take a closer look at bulk copy utility (BCP) or T-SQL bulk insert command. Sorry to disappoint you, Sergei
-
Juergen, If it works for you, why not? No actual DB APIs calls are different. I'd run some tests, because my philosophy is that when you change even a comma in the code, you need to re-run your test suites. Sergei
-
Juergen, Did you try the latest OTL 4.0.469? OTL 4.0.360 is out of date. Sergei
-
Paul, No, OTL as it stands right now, doesn't support connecting to the database using Oracle Wallets. However, otl_connect class has a constructor / logon function that accept raw OCI resources allocated externally. So, you can implement an Oracle Wallet based connector / extender and use it with OTL without touching the rest of your OTL based code. Sergei
-
Maab, What C++ compiler are you using? Do you share the same connect objects between threads? Is it possible for you to put together a self-contained code example that reproduces the problem and email it to me? I've never seen this problem before, at least not with Visual C++ and MS SQL Server, and nobody else reported it (you're the first). Sergei
-
Natalia, I tried to set SQL_COPT_SS_TXN_ISOLATION connection attribute to SQL_TXN_SS_SNAPSHOT using the actual values that I found in my old sqlncli.h (from MS SQL v11). That worked. I also tried to run "SET TRANSACTION ISOLATION LEVEL SNAPSHOT" command using otl_connect::direct_exec(). That worked as well. It just seems that executing "SET TRANSACTION ISOLATION LEVEL SNAPSHOT" is the cleanest solution, and it doesn't require anything dubious (speaking of long term solutions and dependencies on old...
-
Natalia, I tried to set SQL_COPT_SS_TXN_ISOLATION connection attribute to SQL_TXN_SS_SNAPSHOT using the actual values that I found in my old sqlncli.h (from MS SQL v11). That worked. I also tried to run "SET TRANSACTION ISOLATION LEVEL SNAPSHOT" command using otl_connect::direct_exec(). That worked as well. It just seems that executing "SET TRANSACTION ISOLATION LEVEL SNAPSHOT" is the cleanest solution, and it doesn't require any dubious (speaking of long term solutions and dependencies on old header...
-
Natalia, I can see sqlncli.h file only in the SDK for SQL Server version 11, which is pretty old (MS SQL 2012, maybe?). I'll try to use the same values as the corresponding #define have to see what happens. Sergei
-
There are also table hints: https://docs.microsoft.com/en-us/sql/t-sql/queries/hints-transact-sql-table?view=sql-server-ver15 Sergei
-
Natalia, It doesn't look like ODBC supports it. Take a look at your sql.h. I don't see a #define that points to "SNAPSHOT". Here's #defines in my sql.h in Visual Studio 2019: SQL_TXN_READ_UNCOMMITTED (1), SQL_TXN_READ_COMMITTED (2), SQL_TXN_REPEATABLE_READ (4), SQL_TXN_SERIALIZABLE (8), SQL_TRANSACTION_READ_UNCOMMITTED, SQL_TRANSACTION_READ_COMMITTED, SQL_TRANSACTION_REPEATABLE_READ, SQL_TRANSACTION_SERIALIZABLE. According to this, "snapshot" isolation level may be 32: SQL_TXN_READ_UNCOMMITTED 1...
-
Mike, It would useful if you can put together a self-conatained code example that reproduces the problem and email it to me (skuchin@gmail.com). It's hard to say what's wrong. If I can reproduces on my system, I should be able to say what's wrong and how to fix it. Sergei
-
Michael, What ODBC driver against what version of SQL Server are you using. I expect you'd be using one of these ODBC drivers: ODBC Driver 11/13/17 of SQL Server. If you're supposed to use only he ODBC driver you mentioned in your post, you can try to switch from #define OTL_ODBC_MSSQL_2008 to generic #define OTL_ODBC. Sergei
-
Optimus, You can read the OTL manual in order to learn how to do things: click on "Table of Contents", you'll find code examples there, how to build an executable, etc. I read your second posting. If you want to use "Oracle only", why don't you try what Oracle natively supports. It's called OCCI. It's a "C++ OCI". Cheers, Sergei Kuchin
-
I'll make the change to OTL in the meantime to remove the call to std::uncaught_exception() in flush() when OTL_DESTRUCTORS_DO_NOT_THROW is enabled. I'll publish it as OTl 4.0.456. Check for updates. On Tue, Jun 2, 2020 at 9:04 AM Ahsan Baig mmajbaig@users.sourceforge.net wrote: Hmm. Ok. So we'll try using the OTL_DESTRUCTORS_DO_NOT_THROW flag and remove the call to std::uncaught_exception and will let you know if it does resolve our issue. However, it may take around a week or more for us to know...
-
But the destructor of otl_tmpl_out_stream also sets the in_destruct_flag to 1 which is checked inside the flush function. So that case is already handled. Correct me if I'm wrong. Thanks. in_destruct_flag is only useful if another otl_exception is in progress. If an exception of another type is in flight then the call to std::uncaught_exception() is the only way to determine stack unwinding. On Tue, Jun 2, 2020 at 8:53 AM Ahsan Baig mmajbaig@users.sourceforge.net wrote: But the destructor of otl_tmpl_out_stream...
-
I guess it's safe to remove the call to std::uncaught_exception() in flush() if you use #define OTL_DESTRUCTORS_DO_NOT_THROW. So, question for you: if you can do that in you code base, I can add OTL_DESTRUCTORS_DO_NOT_THROW based conditionally compiled logic to flush(). On Tue, Jun 2, 2020 at 8:41 AM Sergei Kuchin skuchin@users.sourceforge.net wrote: I'll take a closer look at the traces. To answer your why question about the reason of calling std::uncaught_exception the intent was to determine the...
-
I'll take a closer look at the traces. To answer your why question about the reason of calling std::uncaught_exception the intent was to determine the stack unwinding situation. flush() may be called from the stream destructor if the stream buffer is not clean, which may throw another exception in case of a database error and trigger a call to std::terminate(). If you have any suggestions on how the stack unwinding can be determined in a safer way I should be able to implement that. Your situation...
-
Ahsan, Are you using Sun C++? Do you have a reference to a resource (discussion group, manual, etc.) that says anything about std::uncaught_exception() being thread unsafe on Solaris? OTL has been in existence for a long time and nobody reported this problem before. If your version of C++ supports C++17, you can try to enable OTL_CPP_17_ON and see if it makes any difference. It uses std::uncaught_exceptions(). Sergei On Tue, Jun 2, 2020 at 3:23 AM Ahsan Baig mmajbaig@users.sourceforge.net wrote:...
-
Will do.
-
Stefano, I just uploaded what you need. Check out What's New page. Cheers, Sergei
-
Stefano, I just uploaded what you. Check out What's New page. Cheers, Sergei
-
Stefano, Here's an example of fast pace development (https://docs.microsoft.com/en-us/visualstudio/releases/2019/release-notes): What's New in Visual Studio 2019 Visual Studio 2019 Releases November 12, 2019 — Visual Studio 2019 version 16.3.9New release icon November 5, 2019 — Visual Studio 2019 version 16.3.8 October 29, 2019 — Visual Studio 2019 version 16.3.7 October 22, 2019 — Visual Studio 2019 version 16.3.6 October 15, 2019 — Visual Studio 2019 version 16.3.5 October 10, 2019 — Visual Studio...
-
Hi Stefano, How many versions back from the latest OTL do you need to go? OTL only changes as fast as C++ compilers release new versions with breaking changes. I can't afford supporting more than one version of OTL these days. Let's say, your package that you're working on has OTL v-5 from the latest. It means that by the time you release your package, there are C++ compilers in the wild that OTL v-5 doesn't cover. I'm not sure sure if it's better if you simply took the latest OTL of the time of...
-
Stephan, Can you try OTL code example 682 (http://otl.sourceforge.net/otl4_ex682.htm) in your envorrinment to see if you get the following output: f1=1, f2=23:12:12.1234567 f1=2, f2=23:12:12.1234567 f1=3, f2=23:12:12.1234567 f1=4, f2=23:12:12.1234567 f1=5, f2=23:12:12.1234567 f1=6, f2=23:12:12.1234567 f1=7, f2=23:12:12.1234567 f1=8, f2=23:12:12.1234567 f1=9, f2=23:12:12.1234567 f1=10, f2=23:12:12.1234567 If it's not what you expect, feel free to email me at skuchin@gmail.com with more detail. Ideally,...
-
Allen, In Windows, ODBC Driver for PostrgeSQL has "DataSource" / Advanced Options dialog, which has Max Varchar entry field, which is by default 255. Fox Linux, PostgreSQL ODBC driver must have a similar parameter that can be added to one of .ini ODBC files (odbcinst.ini / odbc.ini). I've not looked myself but it should be easy enough to find the manual of PostgreSQL ODBC driver for Linux with examples. I Cheers, Sergei
-
Good question. Relational databases are not "sexy" any more, they are legacy things. Everything is moving fast (a.k.a. agile) these days, too much time goes into just keeping up with all of the releases of C++ compilers (GNU C++, CLANG, Visual C++). If somebody contributes good enough quality code like it happened with support for geometry for Oracle, I'd be happy to integrate that to OTL. Cheers, Sergei
-
Zach, Can you put together a self-contained code example of you're trying to do and email it to me at skuchin at gmail dot com? What I need is a stored procedure with th exact same set of parameters as your original SP, no need for the business logic, maybe a few rows in the result set, which you can hard code / insert into your simulated table. I'm not sure how well recent ODBC drivers for MS SQL Server (I assume that's what you're using since Sybase is too far gone) support named parameters for...
-
Isn't that something? I'll add the reference to the OTL F.A.Q.
-
It would be a "compile time patch" enabled by a #define. Not sure if you're talking about dynamically identifying what type of database you're connecting to. If you have "database drivers" that are built with different OTL #defines, in that case you'll have to add the new #define to enable the patch. The patch would literally map "database ints" to C++ "bigints" (8-bytes). Question: do you use #define OTL_ODBC_MSSQL_2008 for SQL Server, or just plain vanilla #define OTL_ODBC for all database typ...
-
The ODBC driver is doing it wrong. It needs to report an 8-byte int as an "ODBC big int", not as an "ODBC int", which is 4-bytes long. It may be similar to the old debate "is ODBC long int 4-bytes or 8-bytes". I can add a conditionally compiled patch (enabled with a #define) to OTL to map "ODBC int" to "ODBC big int", if that helps.
-
Something isn't right in the ODBC driver. dbtype of 4 is an int the way the driver reports the column data type and it's a signed 4-byte integer. Yet, when OTL binds an array of ints on the C++ side, the driver returns bad integer values. When the "long" override is used, everything works correctly. "long" is a signed 8-byte whole number in 64-bit Linux. "bigint" is also a signed 8-byte whole number, that's perhaps why it works in 64-bit Windows. I wonder if "int" means an signed 8-byte integer in...
-
Natalia, Can you add the following to your code to see how the SQLite ODBC driver describes the data type: otl_column_desc* desc; int desc_len; desc=myotl_stream.describe_select(desc_len); for(int n=0;n<desc_len;++n){ cout<<"========== COLUMN #"<<n+1<<" ==========="<<endl; cout<<"name="<<desc[n].name<<endl; cout<<"dbtype="<<desc[n].dbtype<<endl; cout<<"otl_var_dbtype="<<desc[n].otl_var_dbtype<<endl; cout<<"dbsize="<<desc[n].dbsize<<endl; cout<<"scale="<<desc[n].scale<<endl; cout<<"prec="<<desc[n].prec<<endl;...
-
Juergen, otl_direct_exec_select only applies to ODBC/DB2-CLI, and you're using Oracle (OCI). Here's an excerpt from the manual: OTL 4.0/ODBC and OTL 4.0/DB2-CLI define the following global constant to be used with this constructor: otl_explicit_select -- to indicate that if the stream is a SELECT statement then the statement is a simple SELECT. otl_implicit_select -- to indicate that the stream is a stored procedure call that returns one result set. otl_direct_exec_select -- to indicate that if the...
-
Juergen, Can you post (or email me directly at skuchin@gmail.com) SQL statements in the excessive amount of open cursors that you described by using: SELECT a.sql_text sql_text FROM v$sqlarea a, v$open_cursor b WHERE b.sid=<my_session_id> AND a.address=b.address Also, what is CURSOR_SHARING in your database set to (I assumed you used Oracle the way your original post sounded) OTL stream doesn't generate any extra SQL statements based on the inputs. It takes the SQL statement, parses out any bind...
-
Jose, OTL supports different ODBC/DB2-CLI cursor types. Here's an excerpt from the manual (http://otl.sourceforge.net/otl3_connect_class.htm): void set_cursor_type (const int acursor_type=0); This function works in OTL 4.0/ODBC and OTL 4.0/DB2-CLI only, and it has not effect in OTL 4.0/OCIx. Sets the cursor type. Once the cursor type is set, the setting will be propagated to all SELECT statements opened via the otl_stream class in the current connect object. The following cursor types are available...
-
Natalia, I'm glad it worked for you. Can you point me to the Microsoft web page / manual that says that SQLExecDirect() is more efficient than pair SQLPrepare()+SQLExecute()? If you're using OTL_ODBC_MSSQL_2008 for your OTL configuration, it doesn't enable OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE, unless you enabled OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE explicitly. I'm just wondering to be able to understand what you said in your last message... Sergei
-
Natalia, I'm glad it worked for you. Can you point me to the Microsoft web page / manual that says that SQLExecDirect() is more efficient than pair SQLPrepare()+SQLExecute()? If you're using OTL_ODBC_MSSQL_2008 for your OTL configuration, it doesn't enable OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE, unless you anabled OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE explicitly. I'm just wondering to be able to understand what you said in your last message... Sergei
-
Natalia, I'm glad it worked for you. Can you point me to the Microsoft web page / manual that says that SQLExecDirect() is more efficient than pair SQLPrepare()+SQLExecute()? If you're using OTL_ODBC_MSSQL_2008 for your OTL configuration, it doesn't enable OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE, unless you anabled OTL_ODBC_SELECT_STM_EXECUTE_BEFORE_DESCRIBE explicitly. I'm just wondering to be able to understand what you said in you last message... Sergei
-
Natalia, I forgot to plug in an obvious hole in the logic and it was reported less than 24 hours later. So, if you've not started using OTL 4.0.413, I' d recommend you to download OTL 4.0.414. Sergei
-
Natalia, I forgot to plague in an obvious hole in the logic and it was reported less than 24 hours later. So, if you've not started using OTL 4.0.413, I' d recommend you to download OTL 4.0.414. Sergei
-
Hi Natalia, It's the opposite of what OTL was optimized for. However, I'll take a look, and if I can parameterize it in a simple enough way (if it doesn't take too much time and doesn't break things in subtle ways), I'll implement what you need, and let you know. Cheers, Sergei
-
Natalia, I just uploaded OTL 4.0.413 that implements the following new value for "implicit_select" parameter in otl_stream::open() and otl_stream::otl_stream() for ODBC/ DB2-CLI: otl_direct_exec_select. See What's New at the OTL Web site for more detail. Cheers, Sergei
-
Hi Natalia, It's the opposite what OTL was optimized for. However, I'll take a look, and if I can parameterize it in a simple enough way (if it doesn't take too much time and doesn't break things in subtle ways), I'll implement what you need, and let you know. Cheers, Sergei
-
Hi Superantimonitor, These may be appropriate: http://otl.sourceforge.net/otl3_ex33.htm http://otl.sourceforge.net/otl4_ex163.htm http://otl.sourceforge.net/otl4_ex263.htm They are code examples from http://otl.sourceforge.net/otl4_mssql_examples.htm Also, check out the F.A.Q. for SQL Server limitations and specificities: http://otl.sourceforge.net/otl3_faq.htm SQL Server is different enough from Oracle. It's like swallowing a bitter pill after eating a candy :-) Cheers, Sergei
-
Misread your post and though you used Sybase. If you're using OTL in Linux against MS SQL Server, there is a native ODBC driver for SQL Server from Microsoft. The link to download the driver is mentioned in the OTL FAQ: OTL: Does OTL work with MS SQL Server and Linux? Yes, it does. The old way is to use the FreeTDS ODBC driver. FreeTDS lacks any support for the TDS protocol for MS SQL Server 2005 (and higher). A few years back, Microsoft released a 64-bit SNAC / ODBC driver for Linux. It's not Open...
-
Sybase is this case is similar to MS SQL, and the question is answered in the OTL FAQ: Q. OTL: Why my T-SQL stored procedure that has INSERT or UPDATE or DELETE statements doesn't work with OTL? INSERT/UPDATE/DELETE statеments inside T-SQL stored procedures generate row [processed] counts (RPC). The RPCs get returned back to the caller, in this case to the OTL/C++ code, as separate result sets via MS SQL Native Client (SNAC). OTL doesn't handle such RPC result sets, so they need to be turned off...
-
SET NOCOUNT ON
-
ODBC drivers are somewhat different. There are no code examples to show how to use...
-
Gennady, The message definitely says that the ODBC DSN wasn't found. Not sure why....
-
Paul, You may want to put together a self-contained code example of what you're trying...
-
Alex, Make sure that you're using a SQL Server ODBc driver / SQL Native Client that...
-
If you can do an ODBC trace (ODBC driver manager, Tracing tab) on the slow connect,...
-
Hi, Michael, Maybe you're missing a call to otl_connect::commit()? By default, otl_connect::rlogon()...
-
Is it possible to set the driver which should be used in the connection string? Eg....
-
Tom, Got your long email. My gmail address is at the bottom of the OTL Web pages....
-
Tom, Email the error message to my gmail address directly, I'll try to help you....
-
Juergen, I received an email from you, but I get an email delivery error when I try...
-
Juergen, There may be some SQL Client settings for this kind of interruptions / fail-overs....
-
Juergen, I don't watch the OTL forum that closely these days. Here's a quick code...
-
Juergen, I don't watch the OTL that closely these days. Here's quick code example...
-
Hemny, Can you email me (skuchin at gmail dot com) your Informix header files and...
-
Tanz, otl_connect doesn't auto commit on logoff(), but Oracle commits the last transaction...
-
Let's take this discussion offline. Send an email to my gmail account (can be found...
-
Let's take this discussion offline. Send me an email to my gmail account (can be...
-
Bogdan, Strangely enough, my last reply didn't get through to the OTL forum. I said...
-
Yes, it relies on the connected flag. If you find any flaw in the logic, let me know,...
-
logoff() is called in here virtual ~otl_tmpl_connect() OTL_THROWS_OTL_EXCEPTION {...
-
Correct.
-
define OTL_STL is needed when using std::string, otl_input_iterator, and otl_output_iterator....
-
Hi Bogdan, I think you found a documentation bug. I extended OTL stream pooling to...
-
Rustam, A fix for this specific problem is available (OTL 4.0.319). You may want...
-
Hi Martin, OTL wasn't very popular among the boost crowd. They said it was crap,...
-
Rustam, Is it something that you can reproduce in a self-contained code example?...
-
Hi, Samira, Are you talking about a plain C compiler (c99)? If so, OTL is a C++ library....
-
Robert, RIIA idiom is not complete: it recommends closing resources when objects...
-
Thomas, I checked with the latest patch level of xlC 10, your test program shows...
-
Thanks for the info. On Thu, Feb 20, 2014 at 5:54 AM, Thomas Munro tdmunro@users.sf.net...
-
Thomas, Was the problem in 32-bit or 64-bit mode? Thanks, Sergei Sergei Kuchin wrote:...
-
I'll add #define OTL_UNCAUGHT_EXCEPTION_OFF for you to be able to turn the feature...
-
There are other versions of xlC, like 10, 12, and there are different patch levels...
-
You only need max(length(CLOB)) for the same SELECT in order to adjust the char[XXX]...
-
Robert. BTW, did you call set_lob_prefetch_size() before opening the otl_stream?...
-
Robert, Maybe you can get the CLOB lengths first, readjust your SELECT that has char[XXX]...
-
Robert, I implemented otl_connect::set_lob_prefetch_size() (see OTL 4.0.316), which...
-
Robert, I'll take a look at what I can do. It just looks that it should be easy enough...
-
Forgot to include a quote from the OCI article that Robert referred to: Performance...
-
Robert, You can use char[XXX], or raw[XXX] to override the default data type mapping...
-
Let's take this discussion "off line". Can you send me an email to skuchin at gmail...
-
It would be cleaner for you to use the following variant of the rlogon method: otl_connect::rlogon(OCIEnv...
-
Ezequiel, I converted OCIStmtPrepare() calls to OCIStmtPrepare2() calls. Check out...
-
otl_stream has a built-in function get_rpc(): code example otl_stream i(50, // buffer...
-
Ezequiel, I'm aware of this feature. It's a matter of finding enough time to add...
-
Take a look at: http://otl.sourceforge.net/otl3.htm#toc, Declaration of bind variables....
-
Hi Meir, Here's an excerpt from the "What's new" page: 21-Jul-2013, (New in OTL 4.0.298):...
1