|
Thanks for the reply, this is the code - it crashes on the last line. It also crashes when filling a DataTable.
SqlCeConnection conn = new SqlCeConnection( connectionString );
conn.Open();
SqlCeCommand selectCmd = conn.CreateCommand();
selectCmd.CommandText = @"Select Value From MyTable where ColumnOne=101";
SqlCeDataAdapter adp = new SqlCeDataAdapter( selectCmd );
System.Data.DataSet ds = new System.Data.DataSet();
adp.Fill( ds );
|
|
|
|
|
Your code looks good and should execute without problems.
Is any stack trace of this error available?
(The complete trace from adp.Fill(...) to the access violation would be nice)Greetings
Covean
|
|
|
|
|
Respected,
i want one requirement we maintain the web site,so every 1 month my company newsletter will be send to the mailID's automatically ,requirement is one textbox and one button ,when u subbmit u r mailid the newsletter will send that time and every 1 month this is my requirement plse any body helpme sample application
Thanks,
|
|
|
|
|
Try rentacoder, they sell their services really cheap, people here are trying to learn to rather than asking for codez.
If you are going to do the job look into SQL server scheduled jobs and database email.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
I know it is possible to use SSIS and other tools to translate data from one schema to another but does anyone know of any well-known patterns or code examples of doing this?
One wrinkle to the problem is that the original constraints, foreign keys, unique keys, etc., were not enforced at the database level and so translation has to consider this. In other words, fix things up as best as possible by doing things like inserting rows into parent tables when foreign keys are missing.
The data in question isn't very large; a max of around 200MB.
The spec would like the solution to be in two phases, one to translate into the new schema with data written out to the file system and another phase where the data is loaded into the new database, MS SQL Server.
Unless there is a free tool that can do this and do it well, the solution will need to be hand coded. Someone had already looked at using a mapping tool with poor results.
|
|
|
|
|
Having done this job a number of times over the years, I doubt there could be a tool to do this. So many of the decisions are a judgement call that an automated tool could not cut it.
treefirmy wrote: doing things like inserting rows into parent tables
This typical example - insert the parent or delete the child - is typical.
These type of jobs are where we really earn our money, if they ever automate it I'll be astonished but would like to meet the AI that does it.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Thanks for the info.
Are there any online articles, etc. that might be helpful for me to read?
Since you have done this many times, is there any wisdom you might impart? This will be my first time. What things should I expect to run into other than memory issues?
Unfortunately, the data is coming in as a DataSet. Is that even reasonable? I’ve been told that DataSet is space inefficient. Are there other problems with DataSet to watch out for?
My idea was to work with the data in place:
1. Scrub the data so that it passes the value constraints of the new schema (rules come from an XML file)
2. Rename tables and columns
3. Create new tables, filling them with columns and rows of data (not sure this will even work - maybe have to create a new set and move data into it)
4. Fix up foreign key constraints
5. Either drop data for rows that have duplicate primary keys or somehow, make them unique. Both of these options seem very expensive time-wise.
6. Binary serialize the data to a file
For the upload, I was going to use the BulkCopy class. I considered just writing out table data in a format that bcp.exe could understand but bcp.exe only allows 263 rows. (Wow, that limit is weird, http://msdn.microsoft.com/en-us/library/ms162802.aspx).
|
|
|
|
|
Whenever I do this I do it at the database level, I would not even consider doing through any other interface. My usual process is something like:
Take a copy of the production DB
Create a target DB
Create as much of the data structure as you understand based on the current crap and what the business wants to do
Starting with the static tables (Country, City, ###Type etc) and script them into the new database
When all the static stuff is over take a backup of your target DB - this is your new start point
Now start with the ugly stuff, working through your data structures. Create additional tables to hold any new records created to support the new data integrity.
EVERY step requires a script, a reset script and an endless acceptance that NOTHING is final till the next start point. If you get a structure in and are happy with it take more backups as a new point. Always be prepared to trash a start point if you get it wrong.
NEVER accept that a a prior point is sacrosanct if it is wrong scrap it and fix the problem - the other guy probably didn't do this.
If you do a single action in the entire process that is not scripted and repeatable you are screwed and should quit now, before you start. Do not let your PM/boss push you into accepting a wrong decision.
Once you have the whole thing completed script out your target database. Create a new database with the script and run your transfer script using the latest production data, now go fix the NEW problems raised by the latest data, this is a diminishing return problem and needs to be repeated till it... well diminishes.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Thanks for the info – that seems like a sane way to go.
Of course, things are a little more involved as we don’t have the actual data that will be fed to the tool. The tool is supposed to run on customer data, possibly 1000’s of customers with 10-100,000 ‘databases’ each. ‘database’ == ISAM database. The company of course has sample ‘databases’ and can create ‘databases’ but it does not have the actual data that this tool will be fed.
Using a script seems like a much better way to go as the script could be modified by a DB admin at the customer site whereas doing it in code means rewriting/refactoring and recompiling (well unless they could somehow jam the change into the XML file – yeah, I know doubtful).
I’m naïve about SQL scripting, how would one go about actually stuffing the ISAM data into the new SQL server tables? I’ve only ever written C# code to call stored procs. Is this what you mean?
The company already tried to have this project done offshore with terrible results (they completely ignored memory considerations). Unfortunately, I’m not sure I would be able to convince them to go the script way as they seem to want this done in code.
Hopefully this all makes some sort of sense – I’m not even close to being DB savvy.
|
|
|
|
|
treefirmy wrote: I’m not even close to being DB savvy.
The you are in serious trouble. I would not attempt this without the services of a highly experienced and skilled data specialist. As the company has already found out.
This sounds like it is related to a previous post I read a couple of weeks ago with "1000s" of databases. I think your company needs to totally rethink your data structure/storage/delivery processes. 1st question I would ask is do you own the data, 2nd is do you have a mandate to change.
If either of these is no then stop now, the company is doing a half assed job AGAIN.
Or are you writing an interface between the crap data and something else, if so what is the else?Never underestimate the power of human stupidity
RAH
|
|
|
|
|
The customer owns the data in real terms. The data is only supposed to be changed by the older version of the application so it’s not as if it is an open database where others are allowed to change it.
(On the 1000s of 'databases', my PM, thank goodness, understands that this is not the right way to go and will 'force' the developers to use an extra DB table so there will be just one DB.)
On the mandate issue, I know the new application is going to go to SQL Server no matter what. Well, unless the entire scheme is completely unworkable and there is a vast amount of evidence to show the critical problems. If this is true, I can probably kiss my job goodbye.
No this won't be just an interface; it’s supposed to literally be a scrub, translation, write data to disk and then upload with the tool to the new SQL Server DB. Scrubing and translating data makes me nervous - if I were a customer and I knew this was happening, I would have serious reservations.
What’s the yikes factor given the information above? High I’m guessing  .No one ever lost a fortune by underestimating the intelligence of the American public.
|
|
|
|
|
So my 2nd post still applies. You cannot get the original data but have the current schema. I would create a sql database/table to take current customers data and just grab a substantial set of data to work with. I would then treat that as the source database. Work with that database to create your new data structures.
I am presuming there is 1 table coming from the source system with a humongous amount redundant data in it! If this is the case I usually write a script to move the source FILE into a data structure. I never do the transformation in the data transfer tool eg get the data into the database then clean it up.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Excellent.
How do I convince my PM that this is the way to go? What rationale should be used? (My PM has managed enterprise type data migration but only with SSIS/tools. I think some of the other developers managed the offshore team and either guided them to the scrub in code scheme or at least allowed it.)
Keep in mind this, scrub/translate in SQL Server, might be a very hard sell as I'm not experienced in writing SQL scripts, and have written C++ and C# code for 10+ yrs.
Thanks so much for the information.
p.s. There are a number of files which make up the ISAM database.
|
|
|
|
|
treefirmy wrote: scrub/translate in SQL Server, might be a very hard sell as I'm not experienced in writing SQL
My recommendation is based on my experience and I'm afraid I DO have extensive experience in SQL. Seriously, I would recommend getting in a consultant to do this job, it is not trivial and screwing it up will (and already has once) totally ruin any app built on it!
The main reason not to use SSIS for the transformation is my own lack of skill with the tool, I find it particularly difficult to audit a process that happens on the server in what is essentially a black box. Another is that we recently did some comparisons between using SSIS to transform the data and stored procs and the procs were 3 times faster than SSIS. When you process over 2k files per day it is important.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
I know I'm asking a lot and you should be paid for this kind of info. So thanks again.
Don't get me wrong, if it were up to me I would more than follow your guidance. The only issue here is that I'm going to be a mouthpiece for your ideas and I don't have much standing in the company. I'm a new hire.
Given that, what concrete examples/reasons can I give my PM to convince him of the folly of the current approach and the correctness of the 'move data to SQL Server and then do translation into the new schema' strategy?
|
|
|
|
|
You've recieved some excellent info from Mycroft. And I agree with him fully.
I'd just like to add that memory considerations is a major reason to do the job on the database using SQL scripts.
A database is optimised for handling large sets of data."When did ignorance become a point of view" - Dilbert
|
|
|
|
|
I am trying to call Oracle function via Linked Server, while executing it return the error. Can some one tell what is going wrong with this code, keep in mind that the parameter passed are of exactly same type defined in Oracle function except the date type parameter which I am passing as string because there is no equivalent.
-- returned error --
OLE DB provider "MSDAORA" for linked server "ERPUDEV" returned message "ORA-06550: line 1, column 18:
PLS-00382: expression is of wrong type
ORA-06550: line 1, column 7:
PL/SQL: Statement ignored
-- calling part --
DECLARE @source_code VARCHAR(30)
DECLARE @source_lineid INT
DECLARE @feeder_itemcode VARCHAR(40)
DECLARE @sales_price INT
DECLARE @fdr_org_code VARCHAR(10)
DECLARE @fdr_subinv_code VARCHAR(10)
DECLARE @txn_qty INT
DECLARE @txn_date VARCHAR(30)
DECLARE @feeder_txntype VARCHAR(30)
DECLARE @fdr_subinvcode VARCHAR(10)
DECLARE @fdr_trnsubinvcode VARCHAR(10)
DECLARE @item_desc VARCHAR(10)
DECLARE @ErrMsg VARCHAR(100)
DECLARE @RetVal INT
SET @source_code = 'SSP-SALES'
SET @source_lineid = 8728
SET @txn_no = '15059'
SET @txn_no2 = '29193'
SET @feeder_itemcode = 'SSP1032'
SET @sales_price = 12
SET @fdr_org_code = 'S'
SET @fdr_subinvcode = 'S-SALES'
SET @txn_qty = 1
SET @txn_date = '13/MAR/2010'
SET @feeder_txntype = 'S-SALES_ISSUE'
SET @fdr_trnsubinvcode = ''
EXEC ('BEGIN ? := xx_eginv_f_dfs_erp_insert(?,?,?,?,?,?,?,?,?,?,?,?);END;', @RetVal,@source_code, @source_lineid, @txn_no, @feeder_itemcode, @sales_price,@fdr_org_code, @fdr_subinvcode, @txn_qty, @txn_date, @feeder_txntype, @fdr_trnsubinvcode,@ErrMsg OUTPUT) AT ERPUDEVThe Phantom.
modified on Sunday, March 14, 2010 8:19 AM
|
|
|
|
|
So have you tried different date formats, I would try 'yyyy-mm-dd' as the second most unambiguous format, you have tried the first, Oracle may not like characters in the data and spits the dummy.
Note this is not a solution, just another avenue to try. I have not used Oracle in a decade or more.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
Thanks for the comments.
A little more to understand the problem, I'v found that the error message was because of the returned type boolean (@RetVal).
-- oracle implementation
--======================
CREATE OR REPLACE FUNCTION TEST_ERP_INSERT (
P_SOURCE_CODE IN VARCHAR2
, P_SOURCE_LINE_ID IN NUMBER
, P_TRANSACTION_NUMBER IN VARCHAR2
, P_FEEDER_ITEM_CODE IN VARCHAR2
, P_SALES_PRICE IN NUMBER
, P_FEEDER_ORGANIZATION_CODE IN VARCHAR2
, P_FEEDER_SUBINVENTORY_CODE IN VARCHAR2
, P_TRANSACTION_QUANTITY IN NUMBER
, P_TRANSACTION_DATE IN VARCHAR2
, P_FEEDER_TRANSACTION_TYPE IN VARCHAR2
, P_FEEDER_TRANSFER_SUBINV_CODE IN VARCHAR2
, p_errmesg OUT VARCHAR2
)
RETURN BOOLEAN
IS
BEGIN
RETURN 1;
--
EXCEPTION
--
WHEN OTHERS
THEN
--
p_errmesg := 'Error:'
|| SQLERRM;
--
RETURN 0;
END;
-- calling from sql server
--========================
DECLARE @ErrMsg VARCHAR(1000)
DECLARE @RetVal SMALLINT
EXEC ('BEGIN ? := TEST_ERP_INSERT(?,?,?,?,?,?,?,?,?,?,?,?);END;', @RetVal,'SSP-SALES', 8728, '15059', 'SSP1032', 12,'S', 'S-SALES', 1, '13/MAR/2010', 'S-SALES_ISSUE', '', @ErrMsg OUTPUT) AT ERPUSTG;
Any idea? I can't change the returned type because its a different system.The Phantom.
|
|
|
|
|
Sample datamodel for marketing promotion
|
|
|
|
|
And your question is?
I will take mercy on you and supply this link [^]to a message on this page.Never underestimate the power of human stupidity
RAH
|
|
|
|
|
| Look Here[^] I don't speak Idiot - please talk slowly and clearly
'This space for rent'
Driven to the arms of Heineken by the wife
|
|
|
|
|
Hi,
I Wonder if it possible to retrieve from an SQL server a recored plus its 5 recored before anf after.
for example: When I search a product with serial-number "100" i would like to retrieve 11 recorods from serial-number "095" until "105", of course after a "ORDER BY" statement of the serial-number field
|
|
|
|
|
???
WHERE ID BETWEEN 95 AND 105
depending on the database the limit values will or will not be included (then change to 94 AND 106)
|
|
|
|
|
where serial-number between (100-5) and (100+5)
order by seriual-number
Does that do the trick? Chris Meech
I am Canadian. [heard in a local bar]
In theory there is no difference between theory and practice. In practice there is. [Yogi Berra]
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
