In general, I say replace only those chars that you the algorithm requires you to replace, leave all others alone.

Otherwise, say 6 months from now, they decide to convert it...

maybe something like this

select ic.NumericValue, xml_string_agg.string
from #IntChange ic
     cross apply (select stuff((select ''+v.repl
         ...

I do.

Similarly, if, say, I were substituting letters l and 0 to prevent ambiguity, I'd simply replace just those, not 'A' with 'A', 'B' with 'B', etc.

What about a potential...

Your code also has zeros just disappear completely from your results.  It's never explicitly stated that zeros can't appear in these numbers.  Just in case they do, I...

The...

Try different clustering for the table.  If that doesn't help enough by itself, then you can also look into partitioning the table.

The key thing for determining the best clustered index...

• This reply was modified 4 years, 2 months ago by ScottPletcher. Reason: Added "in batches" to make it clear what I intended there

maybe something like this

select ic.NumericValue, 
       stuff((select '' + v.repl
         ...

IF OBJECT_ID('tempdb.dbo.#translations') IS NOT NULL
    DROP TABLE #translations
CREATE TABLE #translations (
    from_char char(1) NOT NULL PRIMARY KEY,
    to_char...

Yep, doing this yourself will be very complex.

I would let SQL itself determine the type(s) in your situation.  This will involve an extra full load of the data.  Make every...

I second the "nonwork_days" table, but I would make sure to design the reason code table well.

You would want to make sure that the reason code table has a...

I very strongly prefer a separate "nonwork_days" table.  The table needs nothing only the date and a tinyint reason code for why that date is a nonwork day.

The big problem...

Partitioning often serendipitously helps with performance.  That's because people will create a different clustered index to help with the partitioning process and that new clustered index actually does help the...

Again, how do you search the table?  That is, what are the common WHERE clauses that you use on the table?  That is what should control the clustering.

If you (almost)...

How do you search the table?  It's likely you would get better performance using partitioning based on how you search the table, esp. if you can partition / cluster first...

My recommendation  would be to read the documentation for BULK INSERT, which is located at the following URL:

https://docs.microsoft.com/en-us/sql/t-sql/statements/bulk-insert-transact-sql?view=sql-server-ver15

The two keywords to search for in that article are ERRORFILE...