--- name: "DBIx::Class::Storage::DBI::Sybase::ASE" section: "3pm" platform: "linux" package: "libdbix-class-perl" version: "0.082844-1" release: "Debian 13 (Trixie)" platforms: ["linux"] url: "https://new.carta.tech/DBIx::Class::Storage::DBI::Sybase::ASE" --- # DBIx::Class::Storage::DBI::Sybase::ASE(3pm) Sybase ASE SQL Server support for DBIx::Class ## NAME DBIx::Class::Storage::DBI::Sybase::ASE - Sybase ASE SQL Server support for DBIx::Class ## SYNOPSIS This subclass supports DBD::Sybase for real (non-Microsoft) Sybase databases. ## DESCRIPTION If your version of Sybase does not support placeholders, then your storage will be reblessed to DBIx::Class::Storage::DBI::Sybase::ASE::NoBindVars. You can also enable that driver explicitly, see the documentation for more details. With this driver there is unfortunately no way to get the `"last_insert_id"` without doing a `"SELECT MAX(col)"`. This is done safely in a transaction (locking the table.) See "INSERTS WITH PLACEHOLDERS". A recommended connect\_info setting: ``` on_connect_call => [['datetime_setup'], ['blob_setup', log_on_update => 0]] ``` ## METHODS ### connect\_call\_blob\_setup Used as: ``` on_connect_call => [ [ 'blob_setup', log_on_update => 0 ] ] ``` Does `"$dbh->{syb_binary_images} = 1;"` to return `"IMAGE"` data as raw binary instead of as a hex string. Recommended. Also sets the `"log_on_update"` value for blob write operations. The default is `1`, but `0` is better if your database is configured for it. See "Handling IMAGE/TEXT data with **syb\_ct\_get\_data()**/**syb\_ct\_send\_data()**" in DBD::Sybase. ### connect\_call\_datetime\_setup Used as: ``` on_connect_call => 'datetime_setup' ``` In connect\_info to set: ``` $dbh->syb_date_fmt('ISO_strict'); # output fmt: 2004-08-21T14:36:48.080Z $dbh->do('set dateformat mdy'); # input fmt: 08/13/1979 18:08:55.080 ``` This works for both `"DATETIME"` and `"SMALLDATETIME"` columns, note that `"SMALLDATETIME"` columns only have minute precision. ## Schema::Loader Support As of version `0.05000`, DBIx::Class::Schema::Loader should work well with most versions of Sybase ASE. ## FreeTDS This driver supports DBD::Sybase compiled against FreeTDS (\) to the best of our ability, however it is recommended that you recompile DBD::Sybase against the Sybase Open Client libraries. They are a part of the Sybase ASE distribution: The Open Client FAQ is here: \. Sybase ASE for Linux (which comes with the Open Client libraries) may be downloaded here: \. To see if you're using FreeTDS run: ``` perl -MDBI -le 'my $dbh = DBI->connect($dsn, $user, $pass); print $dbh->{syb_oc_version}' ``` It is recommended to set `"tds version"` for your ASE server to `5.0` in `"/etc/freetds/freetds.conf"`. Some versions or configurations of the libraries involved will not support placeholders, in which case the storage will be reblessed to DBIx::Class::Storage::DBI::Sybase::ASE::NoBindVars. In some configurations, placeholders will work but will throw implicit type conversion errors for anything that's not expecting a string. In such a case, the `"auto_cast"` option from DBIx::Class::Storage::DBI::AutoCast is automatically set, which you may enable on connection with connect\_call\_set\_auto\_cast. The type info for the `"CAST"`s is taken from the "data\_type" in DBIx::Class::ResultSource definitions in your Result classes, and are mapped to a Sybase type (if it isn't already) using a mapping based on SQL::Translator. In other configurations, placeholders will work just as they do with the Sybase Open Client libraries. Inserts or updates of TEXT/IMAGE columns will **NOT** work with FreeTDS. ## INSERTS WITH PLACEHOLDERS With placeholders enabled, inserts are done in a transaction so that there are no concurrency issues with getting the inserted identity value using `"SELECT MAX(col)"`, which is the only way to get the `"IDENTITY"` value in this mode. In addition, they are done on a separate connection so that it's possible to have active cursors when doing an insert. When using `"DBIx::Class::Storage::DBI::Sybase::ASE::NoBindVars"` transactions are unnecessary and not used, as there are no concurrency issues with `SELECT` `@@IDENTITY` which is a session variable. ## TRANSACTIONS Due to limitations of the TDS protocol and DBD::Sybase, you cannot begin a transaction while there are active cursors, nor can you use multiple active cursors within a transaction. An active cursor is, for example, a ResultSet that has been executed using `"next"` or `"first"` but has not been exhausted or reset. For example, this will not work: ``` $schema->txn_do(sub { my $rs = $schema->resultset('Book'); while (my $result = $rs->next) { $schema->resultset('MetaData')->create({ book_id => $result->id, ... }); } }); ``` This won't either: ``` my $first_row = $large_rs->first; $schema->txn_do(sub { ... }); ``` Transactions done for inserts in `"AutoCommit"` mode when placeholders are in use are not affected, as they are done on an extra database handle. Some workarounds: - • use DBIx::Class::Storage::DBI::Replicated - • connect another Schema - • load the data from your cursor with "all" in DBIx::Class::ResultSet ## MAXIMUM CONNECTIONS The TDS protocol makes separate connections to the server for active statements in the background. By default the number of such connections is limited to 25, on both the client side and the server side. This is a bit too low for a complex DBIx::Class application, so on connection the client side setting is set to `256` (see "maxConnect" in DBD::Sybase.) You can override it to whatever setting you like in the DSN. See \ for information on changing the setting on the server side. ## DATES See "connect\_call\_datetime\_setup" to setup date formats for DBIx::Class::InflateColumn::DateTime. ## LIMITED QUERIES Because ASE does not have a good way to limit results in SQL that works for all types of queries, the limit dialect is set to GenericSubQ. Fortunately, ASE and DBD::Sybase support cursors properly, so when GenericSubQ is too slow you can use the software\_limit DBIx::Class::ResultSet attribute to simulate limited queries by skipping over records. ## TEXT/IMAGE COLUMNS DBD::Sybase compiled with FreeTDS will **NOT** allow you to insert or update `"TEXT/IMAGE"` columns. Setting `"$dbh->{LongReadLen}"` will also not work with FreeTDS use either: ``` $schema->storage->dbh->do("SET TEXTSIZE $bytes"); ``` or ``` $schema->storage->set_textsize($bytes); ``` instead. However, the `"LongReadLen"` you pass in connect\_info is used to execute the equivalent `"SET TEXTSIZE"` command on connection. See "connect\_call\_blob\_setup" for a connect\_info setting you need to work with `"IMAGE"` columns. ## BULK API The experimental DBD::Sybase Bulk API support is used for populate in **void** context, in a transaction on a separate connection. To use this feature effectively, use a large number of rows for each populate call, eg.: ``` while (my $rows = $data_source->get_100_rows()) { $rs->populate($rows); } ``` **NOTE:** the add\_columns calls in your `"Result"` classes **must** list columns in database order for this to work. Also, you may have to unset the `"LANG"` environment variable before loading your app, as `"BCP -Y"` is not yet supported in DBD::Sybase . When inserting IMAGE columns using this method, you'll need to use "connect\_call\_blob\_setup" as well. ## COMPUTED COLUMNS If you have columns such as: ``` created_dtm AS getdate() ``` represent them in your Result classes as: ``` created_dtm => { data_type => undef, default_value => \'getdate()', is_nullable => 0, inflate_datetime => 1, } ``` The `"data_type"` must exist and must be `"undef"`. Then empty inserts will work on tables with such columns. ## TIMESTAMP COLUMNS `"timestamp"` columns in Sybase ASE are not really timestamps, see: \. They should be defined in your Result classes as: ``` ts => { data_type => 'timestamp', is_nullable => 0, inflate_datetime => 0, } ``` The `"\> is necessary if you use DBIx::Class::InflateColumn::DateTime, and most people do, and still want to be able to read these values. The values will come back as hexadecimal. ## TODO - • Transitions to AutoCommit=0 (starting a transaction) mode by exhausting any active cursors, using eager cursors. - • Real limits and limited counts using stored procedures deployed on startup. - • Blob update with a LIKE query on a blob, without invalidating the WHERE condition. - • bulk\_insert using prepare\_cached (see comments.) ## FURTHER QUESTIONS? Check the list of additional DBIC resources. ## COPYRIGHT AND LICENSE This module is free software copyright by the DBIx::Class (DBIC) authors. You can redistribute it and/or modify it under the same terms as the DBIx::Class library.