Documentation Comments

Use this form to comment on this topic. You can also provide any general observations about the Online Documentation, or request that additional information be added in a future release.

Reality V15.0 ()

File Input and Output (DataBasic) (m618705+egfileio.htm)

To	Reality
Version
Topic
Submitted by
Company
Location
Email address
Comment

File Input and Output

File I/O (READ and WRITE statements and their variants) may have an effect on the efficiency of a DataBasic program. Internally these statements have to:

Set a read/update lock on the group into which the item hashes. This may collide with other processes accessing the same file.
Locate the item in the group if it exists.
Copy the item between file space and DataBasic workspace. If accessing many large items, this may cause the workspace to become fragmented and therefore invoke garbage collect.

READV

READV should only be used when only a few attributes are required from a large item and they are not to be written back using WRITEV. In other cases it is more efficient to read the whole item.

WRITEV

This instruction has to read the whole item into DataBasic workspace, replace the relevant attribute, and then write the item back. Therefore if more than one attribute is to be written, it is more efficient to not use WRITEV so that the item is only written once.

MATREAD and MATWRITE

As MATREAD and MATWRITE have to convert between dynamic and dimensioned arrays they take longer to execute, and therefore should only be used where they provide some other benefit to performance (see topic Dynamic Arrays.).

READV of Attribute 0

A READV of attribute 0 allows the existence of an item to be checked without actually reading the item. This is especially useful when accessing items in remote files as no data needs to be copied over the network.