[sub:Metadata-collection]Metadata collection¶

These docs are outdated, we are working on getting them up-to-date for the next release

Remember, that metadata is the data Leksah has collected from all the Haskell code (including .hi files for installed packages) it could reach on your computer.

The initial scan may take a long time (some minutes); when Leksah starts later, it checks only for changes, but does not scan all files again. Metadata collection depends on the local configuration, especially the list of places where Haskell code may be found, which is entered in the preferences. Occasionally, you may find it useful to rebuild the metadata.

Metadata collection can be manually triggered: If you select Configuration -> Update workspace data, the metadata for the current project is collected. This brings the metadata of the current project up-to-date. You can as well press Ctrl-m or hit this symbol in the toolbar:

If you select Configuration -> Rebuild workspace data, the metadata for the current project is rebuild.

If you select Configuration -> Update system data, Leksah checks, if a new library has been installed and then collects metadata for additions.

If you select Configuration -> Rebuild system data, Leksah rebuilds all metadata, which may take a long time. Currently the preferred way to do this is to call leksah-server -rbo +RTS -N2 from the command line. The reason for this is that the server process may allocate a lot of memory during collection.

Background infos¶

The metadata collection itself proceeds different for workspace and system packages:

For workspace packages Leksah just uses the parser without typechecking and maybe .hi files if available.
For system packages Leksah uses .hi files and if sources are available Haddock as a library.

Collection for system packages works as follows:

Packages you installed with cabal from Hackage. If Leksah can’t find sources, it does a cabal unpack in the source directory you specified for this in the preferences (By default .leksah-**/packageSources).
Source files in the folders listed as source folder in the preferences. Leksah looks for all .cabal files it can find below the source folders. Therefore, Leksah collects source information only from “Cabalized” projects (i.e., projects that have a .cabal file). From this information the file source_packages.txt in the .leksah folder is written. If you miss sources for a package in Leksah, consult this file if the source place of the package has been correctly found. You can run this step by typing in a terminal: leksah-server -o (or –sources).
Problems may occur due to preprocessing, header files, language extensions, etc. Error message produced while metadata collection indicate that not all information for a package was found. If the Haddock call doesn’t succeed, Leksah looks if it can download a prebuild metadata package from the server, if this option is selected in the preferences.
The result of metadata collection is stored in the folder .leksah-*.*/metadata in files called *.lkshm. In this folder for every package a metadata file is stored (e.g. binary-0.4.1.lkshm). These files are in binary format. If you want to rebuild just one package you can delete it here and update the system metadata. The files .lkshe specify the base path to sources, if the collection for sources was successfully or if a metdata file could be downloaded for this.

For the workspace packages a different procedure is used.

Metadata is collected from the source directories of the packages you are working on. The results are stored in a per module base in a folder with the package name (e.g .leksah-*.*/metadata/package-*.*.*).
Update happens on a per file base only for changed source files.

[fig:Metadata-Preferences]Metadata Preferences