Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

When a user types a value into a text field (which is bound to a specific column of the database table) and clicks out, the Servoy Application Server issues a SQL update command to the database to modify the selected record. The resulting change is also broadcast|display/DOCS/Data+Broadcasting\ to all connected clients.

...

The fundamental unit of data binding in both the GUI and the API is the Servoy Foundset|display/DOCS/Foundsets+Concepts\ object.

Client Cache

A Servoy client instance keeps track of which database records are in use. This is called the Client Cache and it optimizes performance by reducing the number of queries made to the database. Records are tracked by primary key. The first time the contents of a record are accessed, the Application Server must issue a query to the database on behalf of the client. The values for all of the columns of the record object are held in memory and therefore, subsequent access of the record will not produce anymore queries to the database. The user experience is greatly enhanced as one can browse quickly between forms and among cached records.
A record may fall out of the cache gracefully to conserve memory and is automatically reloaded the next time it is accessed. This happens at the discretion of the client's caching engine, which is highly optimized. Relieved of the burden of managing memory, the developer can focus on more important things.

...

This functionality is provided by default for all Servoy client types. There is nothing that a developer needs to do to enable it. However, the developer may augment the default functionality by capturing the data broadcast event and invoking specific business logic. Also, Servoy's API provides a means to programmatically update client caches in cases where a record is modified outside of any Servoy Client session, i.e. from another application.

Data Transactions in Servoy

Data manipulations in Servoy happen inside an in-memory transaction. When a user creates or modifies a record, nothing is committed to the database immediately. The Servoy Client tracks all newly-created and modified records, including which columns have changed, their former and latter values. As a user adds/modifies records, the amount of information stored in the In-Memory transaction accrue until they are committed or rolled back. The duration of this In-Memory transaction can be short or long depending on the client's configurable Auto Save setting.

Auto Save: ON

By default, every Servoy client is started with the Auto Save setting initialized to on/true. This means that the In-Memory transaction is typically very short as changes are committed automatically as the user navigates the client session. Specific actions like clicking in a form's area, navigating to a different form, clicking a button, etc. all trigger a save event. Auto Save is ideal for situations where the user is intended to be able to make edits freely.

Auto Save: OFF

The developer may optionally set the Auto Save setting to off/false. This means that the length of the In-Memory transaction is controlled by the developer. As changes accrue, they are never committed until the developer programmatically invokes a save event. It is ideal to disable Auto Save for scenarios where the user is intended to perform edits in a controlled situation where a group of edits may be saved or rolled back all together.

The Auto Save setting can be programmatically changed throughout the duration of the client session to accommodate different modes for different editing scenarios.

The Anatomy of the Transaction

Servoy provides a robust data API, giving the developer full access to the In-Memory transaction, which consists of a listing of all record objects that were added or modified. For each of these record objects, there is a listing of every column whose value was changed. For every modified column, there is a reference to the value before and after the edit. The transaction API also allows developers to distinguish between records that are newly-created and do not yet exist in the database, versus records that already exist in the database, but have outstanding edits.

Saving Data Changes

A developer can programmatically issue a save event, causing the contents of the In-Memory transaction to be automatically translated into instructions to insert/update database tables. A developer can optionally invoke a save event for a specific record only, leaving the rest of the transaction unaffected.

If for some reason one or more records were unable to be saved (i.e. due to a back-end database violation, etc.), the transaction will also keep track of Failed Records and their associated errors.

Rolling Back Data Changes

A developer can programmatically issue a command to rollback the contents of the entire In-Memory transaction, causing newly created records to be removed and modified records to be reverted to their state prior to the start of the In-Memory transaction. The developer can optionally choose to rollback changes for a specific record, leaving the rest of the transaction unaffected.

What about Deleting Records?

It is important to note that record deletes are not part of the In-Memory transaction. When a record is deleted, the instructions are sent to the database immediately and the delete cannot be rolled back.

Panel
titleQuestion from Patrick Talbot that this chapter could answer

I remember I was thinking that there were some gray areas when it comes to datasets and foundsets, and generally how Servoy is generating the SQL from these depending on relations etc. I noticed that there were often questions about that in the forum, and no definitive answers of course (but at least pointers to what approach gives you what), and in my company too, people are coming to me quite often with some of the following questions:

  1. the 200 records loaded in a batch, and how to treat it correctly
  2. databaseManager.loadRecords() different flavors and what it does exactly
  3. how Servoy is treating DB Views (no refreshing, need to manually add the PK(s))
  4. calculations and aggregations and how/why they will deteriorate performances (what is the SQL involved especially in case of table/list forms)
  5. about dataBroadcast, how they act on foundsets but not datasets (nor foundsets based on views)
  6. dataBroadcast again: how to refresh you client's data after a batch (processor) update (maybe demonstrate the new headless_client plugin?)
  7. how to use the JSFoundsetUpdater properly, and how to efficiently update a whole foundset in one go
  8. the use of databaseManager.setCreateEmptyFormFoundsets() and what it means
  9. the advantage of using valueLists instead of relations to display related data in table/list forms
  10. how/why Servoy is sometimes creating temporary table to perform some joins
  11. generally how to interpret the performance tab in the server admin

Seems like a lot of topics in one, really, but of course not necessarily all of these topics need to be addressed, or addressed at the same time, but I really think that these are things that need more explanations/clarifications/demonstrations.

...