Zeoslib Portal

Hi All! I've always used IndexFieldNames to sort and it seemingly works very well. I recently started using TZMemTable (very nice btw!) and ran into the issue that the base class for all of the various TZDataSet classes is TZAbstractRODataset and its IndexFieldNames property is protected (and so I'm...

On the oracle utf16 buffer issue: The Oracle data loss is specific to surrogate pair characters in UTF16 mode. I only found it out when doing some "edge case" type testing. I don't suspect it would happen very often "in the wild", but I tend to err on the side of caution and I ha...

no, there currently is no official statement on what Oracle versions we support. I cannot test anything older than Oracle 12c. Since Oracle 9i is kinda ancient, I don't see a reason to support Oracle 8. I'm in complete agreement. I was surprised recently to come across a 9i installation at a custom...

I was recently doing some work in ZDbcOracleMetadata (to add table comments to the UncachedGetTables results) and needed to do some joins and noticed that we use the "newer" join syntax that was added in Oracle 9i in a few places (which is recommended for newer servers and I'm happy to aba...

I've never liked "Unicode" as a character set designator just because it's hard to know exactly what it means. Microsoft (and then Delphi) decided to use it as an alias for UTF-16LE decades ago, but MS reversed that decision a couple years ago and now you can see the much better list in no...

I may be reading the suggestion for Firebird incorrectly, but according to their docs UNICODE_FSS isn't recommended. https://firebirdsql.org/file/documentation/html/en/refdocs/fblangref50/firebird-50-language-reference.html#fblangref50-datatypes-chartypes-unicode Also I believe the "Unicode&quo...

Hi Jan, I believe so. I've tested pretty extensively on that and haven't seen any issues, and I was able to cause it regularly with the old code. The new individual buffers and dynamic arrays for the indicator and ALen variables also allow a few fixes (which I'm currently working on and will submit ...

Unfortunately, the performance part of my fix for the ORA-01406 issue is not correct. I've been testing using strings filled with the "Hwair" character (which is 4 bytes in both UTF8 and UTF16) and my "fix" truncates the data silently in UTF16 mode. The truncation is reported in ...

> As Oracle advises you not to use BasicFile, is it possible that Oracle Cloud even denies it's usage...?

I think that's very possible! No luck with the alter command.

-Mark

Is anything else needed to create a clob that uses basicfile? Here's my create table statement which doesn't return an error, yet creates the clob as securefile. create table bstestbasic ( id number(38) not null primary key, notes clob ) lob (notes) store as basicfile; select table_name, column_name...

Feeling a bit better and got a pull request in. Let me know how it goes.

-Mark

I am, but I've been waylaid by covid this week... Hopefully later in the week I'll be up to it.

-Mark

I was able to tentatively fix this issue by replacing the single large FRowsBuffer with individual buffers for each column item array. I didn't' find any smoking gun in the current code, but my guess is the issue relates to memory alignment. The Oracle OCI docs state: Note:Output buffers must be 2-b...

I ran into an issue with the rounding of nano timestamps a while ago. The suspect code is in ZSysUtils. My test case is: x := RoundNanoFractionTo(123456789, 3); and x := RoundNanoFractionToMillis(123456789); and I think x should be 123000000 but the actual result is 124000000. This tracks back to: c...

Nothing I did seemed to help until I replaced the current FRowsBuffer with my own buffers that I'm storing in each TZSQLVar. I have it working perfectly (so far lol) with UTF16. I'm not quite sure what this means, but whatever is causing the issue appears to happen in procedure TZOracleResultSet.Ope...