CurryInfo: Managing analysis and verification information about Curry packages

This package contains the implementation of a tool to collect and provide analysis and verification information about Curry modules contained in Curry packages. It can be used by other Curry tools which require such information, like a REPL or a Curry Language Server.

Installation

The tool itself can be simply installed by

> git clone https://github.com/curry-language/curry-info-system.git
> cd curry-info-system
> cypm install

This installs the executable curry-info which organizes the generation of and access to various information about Curry module contained in packages.

Basic usage examples

The following command outputs the list of all operations defined in the module Data.List of package base with version 3.4.0:

> curry-info -p base -x 3.4.0 -m Data.List operations

The following commands show the definitions of the type class Ord and signature and operation lines (defined in the Prelude of package base with version 3.4.0), respectively:

> curry-info -p base -x 3.4.0 -m Prelude -c Ord definition
> curry-info -p base -x 3.4.0 -m Prelude -o lines signature
> curry-info -p base -x 3.4.0 -m Prelude -o lines definition

The package and version options can be omitted if the module is part of a package which can be found in the current load path. For instance, the previous commands can be shortened as follows (provided that the Curry system is based base package with version 3.4.0):

> curry-info -m Prelude -c Ord definition
> curry-info -m Prelude -o lines signature
> curry-info -m Prelude -o lines definition

To show results about the operational behavior of operations, curry-info uses other analysis and verification tools which exist for Curry. If they are not installed, you cannot generate and access this information, but only show some basic information about operations, like their signature and definition as shown above.

In order to install the analysis and verification tools currently supported and used by curry-info, run the following commands:

> cypm install cass              # installs executable cass
> cypm install verify-non-fail   # installs executable curry-calltypes

When curry-info is called without options, the list of all packages currently stored in CurryInfo is shown:

> curry-info
packages: abstract-curry abstract-haskell addtypes ...

Web service / CGI mode

If the option --cgi is given, curry-info is executed in CGI mode. In this mode, the parameters are obtained from the value of the environment variable QUERY_STRING which must contain all parameters as URL-encoded strings separated by the character &. The CGI mode is useful to maintain a web server containing the infos about all packages. An example installation of such a server can be accessed at URL

https://cpm.curry-lang.org/curry-info/run.cgi

For instance, the following URL retrieves all versions of the package base:

https://cpm.curry-lang.org/curry-info/run.cgi?-f0&--package=base&versions

The CGI mode and this server is used by the tool cpm-query (see Curry package cpm-query) to query information about Curry entities defined in a module of some existing Curry package.

Installing a CurryInfo Web Application

In order to install a CurryInfo web server to be used as shown above, run

> make install

This compiles the tools and install them in a local directory (see definition of WEBDIR in the Makefile or run make showconfig to see its value). Note that this should be done on the web server or a compatible machine.

After this local installation, run

> make tar

to generate a tar file WEBCURRYINFO.tgz with the necessary contents of WEBDIR. Now copy and unpack it in the desired directory of the web server (e.g., .../curry-info).

Note that the cache of this installed web server is empty. There are two options to fill the cache:

• Copy an existing CurryInfo cache into the directory .curry_info_cache of the web server.

• Updating/generating analysis information for single package versions by

  > cpm-query --remote --generate <PACKAGE> <VERSION>
  

  or updating/generating analysis infos for all (newest compatible) package versions by

  > cpm-manage packagelist | cpm-query --remote --generate --from=-
  

  Note that these commands need to be executed on the host of the CurryInfo web server (see CurryInfo.Options.managerAddrs) since data changes are generally not allowed in CGI mode.

  Furthermore, cpm-query generates a configuration file ~/.cpmqueryrc where one has to set the option curryinfourl to the URL of the CurryInfo web server.

HTML generation

If the option --htmldir=DIR is given, curry-info generates an HTML representation of all entities stored in its cache in the directory DIR. For instance, the representation of the entities stored in the curry-info web service mentioned above can be accessed at URL

https://cpm.curry-lang.org/curry-info/HTML

On the standard CurryInfo web server, these web pages are nightly updated when new package versions arrive.

Documentation

The motivation and ideas of CurryInfo are described in the following paper:

M. Hanus: CurryInfo: Managing Analysis and Verification Information about Curry Packages, Proc. of the 35th International Symposium on Logic-Based Program Synthesis and Transformation (LOPSTR 2025), Springer LNCS 16177, pp. 123-134, 2025, DOI: 10.1007/978-3-032-04848-6_8

Technical notes

• The name of a module entity (operation, type, class) can also be an empty string. In this case, the analysis is performed (if the value of the force option is non-zero) but a result is not shown. This is useful for program analyses which compute at once information about all operations of a module, as the analysis and verification tools cass and verify-non-fail. For instance, the following command generates determinism information for all operations of the prelude:

  > curry-info -f2 -p base -x 3.4.0 -m Prelude -o '' cass-deterministic
  

• CurryInfo is designed as a generic and extensible system to collect and manage information about Curry entities. Thus, it is also possible to add new information requests to CurryInfo.

• The easiest way to use CurryInfo and the information stored in the central CurryInfo server is provided by the tool cpm-query (see Curry package cpm-query). CurryInfo provides also a socket connection with a specific server protocol.

Files

CurryInfo maintains a file cache containing all currently available analysis and verification information. As a default, these files are stored in the directory ~/.curry_info_cache. This path is defined in CurryInfo.Options.defaultCacheRoot but can be changed by the option --cache.

The contents of the cache directory has the following structure:

• checkouts (contains clones of packages checked out by cypm checkout)
  • package1-version1
  • package2-version2
  • ...
• packages
  • pkg
    • pkg.json
    • versions
      • vsn
        • vsn.json
        • modules
          • mod
            • mod.json
            • types
              • type.json
            • classes
              • class.json
            • operations
              • op.json

For instance, the information about the prelude operation foldr1 contained in version 3.4.0 of the base` package is stored in file

.../packages/base/versions/3.4.0/modules/Prelude/operations/foldr1.json

To speed up the access to CASS-requests for all operations, CurryInfo also writes files named ALL_<req>.txt in operations directories. These files contains the CurryTerm output format of these requests for all operations.