-
Notifications
You must be signed in to change notification settings - Fork 17
require.q
This library is inspired by RequireJS and other import systems within JavaScript applications. It allows the loading of libraries into a kdb+ process and manages dependencies between them to aid separation of code into small, reusable components.
We define a library to be a collection of q-script files that are related by a common prefix. In the simplest case, a library can be a single q-script with the name equalling the library name (e.g. log.q
provides the log
library). We also assume that the library name matches the namespace used for that library (e.g. log
library uses .log
namespace).
To allow code within a single library to also be split into separate files, files with the same common prefix are assumed to be the same library. The suffixes used to determine this are: .q
, .k
, .*.q
, .*.k
, .q_
, .*.q_
. Note that with the prefixes .*.q
, .*.k
and .*.q_
, log.q
and, for example, log.extra.q
would be classified as the same library and would both be loaded as part of the log
library.
To start using Require within your application, you must load and initialise the library manually:
q) system "l /path/to/kdb-common/src/require.q"
The library must then be initialised. There are a number of options, depending on the customisation required:
-
.require.init[]
: Uses the current folder as the root to discover files -
.require.init *path*
: Uses the specified path as the root folder to discover files - Setting
.require.location.discovered
: Manually define list of files to use with require
The scanning of files is only done on initialisation, any added or removed files will not be reflected in this list. You should use .require.rescanRoot
to update the list before attempting to load a newly defined library.
By calling .require.init
with no arguments, it will take the current working directory (equivalent of pwd
) as the root folder and scan for all files underneath this folder.
> cd /home/jas/git/kdb-common
> rlwrap $QHOME/l64/q src/require.q
q) .require.init[]
Library root location refreshed [ File Count: 51 ]
q) .require.location.root
`:/home/jas/git/kdb-common
By calling .require.init
with a single folder-path argument, it will use the specified folder as the root folder and scan for all files underneath this folder.
NOTE: It is recommended to pass an absoluate path to .require.init
to allow it to continue working even if the process working directory is later changed (e.g. system "l"
).
> rlwrap $QHOME/l64/q /home/jas/git/kdb-common/src/require.q
/ Equivalent to the "current folder as root" example
q).require.init `$":/home/jas/git/kdb-common"
Library root location refreshed [ File Count: 51 ]
q) .require.location.root
`:/home/jas/git/kdb-common
In the case where there are files from multiple distinct locations on the filesystem that should be exposed to 'require', override the default .require.rescanRoot
function to build the .require.location.discovered
list of files manually.
Notes:
-
.require.i.tree
can be used to descend within a list of folders. -
.require.location.root
will be set to the current working directory regardless of the list of files present, so should be ignored
> rlwrap $QHOME/l64/q /home/jas/git/kdb-common/src/require.q
/ Need kdb-common and kdb-systemd here
.require.rescanRoot:{ .require.location.discovered:raze .require.i.tree each `$(":/home/jas/git/kdb-common"; ":/home/jas/git/kdb-systemd"); };
.require.init[];
Once Require is loaded and initialised, you can then use .require.lib
to load any other library. This function will search for a library with the specified name and load all matching files into the current process. If there is a function *lib*.init
, it will be executed with a single dictionary argument after all the files are loaded.
If you need to perform some other set up or configuration after the files are loaded but before the initialisation function is run, you should use .require.libNoInit
to load the library, run you custom code and then run .require.lib
to initialise it.
If the function *lib*.init
exists, it is called with a single dictionary argument. The dictionary contains the following keys:
-
reinit
: Boolean set to false on first initialisation (via.require.lib
or.require.libForce
) or true if re-initialised via.require.libForce
.require.lib
can also be used to define dependencies between libraries that you write. Use it at the top of your library.
For example, if you need to use .type.isInteger
from the type
library in some new code, you can do the following:
// your-lib.q
/ Loads the "type" library (if not already loaded)
.require.lib `type;
.lib.intSum:{[x;y]
if[(not .type.isInteger x) | not .type.isInteger y];
'"IllegalArgumentException";
];
:x + y;
}
The Require library keeps track of all libraries that are loaded (and initialised). This is used internally to know which libraries to load from disk but can also be used to manually see which libraries are within the process.
The table .require.loadedLibs
provides this detail:
q) .require.loadedLibs
q).require.loadedLibs
lib | loaded loadedTime initExists inited initedTime forced files
-------| ------------------------------------------------------------------------------------------------------------------------------
require| 1 2024.04.17D21:54:46.715152900 1 1 2024.04.17D21:54:46.715163600 0 ()
type | 1 2024.04.17D21:54:46.725174900 1 1 2024.04.17D21:54:46.725710400 0 ,`:./src/type.q
if | 1 2024.04.17D21:54:46.726011100 1 1 2024.04.17D21:54:46.726160900 0 ,`:./src/if.q
...
cron | 1 2024.04.17D21:54:46.777817700 1 1 2024.04.17D21:54:46.778018800 0 ,`:./src/cron.q
There is also the option to force a reload and re-initialisation of a library via .require.libForce
. This can be useful in situations where a library has been modified and can be safely reloaded within a process without having to restart it.
The require
library explicitly logs if a forced reload is performed on an already loaded library and updates the forced
column in .require.loadedLibs
to true.
Notes:
- Any dependent libraries are not reloaded
- If the library is not yet loaded,
.require.libForce
behaves identically to.require.lib
- If the library is loaded but not yet initialised,
.require.libForce
will call the*lib*.init
function withforce
set to false
/ Load 'ns' normally
q) .require.lib`ns;
...
/ Now force a reload
q) .require.libForce`ns;
2023.10.31 17:23:53.848 INFO pid-1875 jas 0 Force reloading library [ Library: ns ] [ Already Loaded: yes ] [ Already Initialised: yes ]
...
q) .require.loadedLibs[`ns]`forced
1b
The following exceptions can be thrown when using the Require library:
-
LibraryDoesNotExistException
: If there are no files that match the library requested -
LibraryLoadException
: If any of the matched files fail to load into the process- A back trace will be printed if running kdb+ 3.5 or later
-
LibraryInitFailedException
: If a initialisation function exists for the library and it does not execute successfully- A back trace will be printed if running kdb+ 3.5 or later
-
UnknownLibraryException
: If you attempt to initialise a library that has not been loaded by Require -
RequireReinitialiseAssertionError
: If a library re-initialisation is attempted, but the re-initialise parameter of.require.i.init
is false
The Require library provides very basic logging for the .log.if
interfaces that all libraries within kdb-common use to print information.
These are defined by .require.i.log
(prints to standard out) and .require.i.logE
(prints to standard error).
These logging functions also support slf4j-style parameterised logging (as does the full logging library) which can be accessed for other use via .require.i.parameterisedLog
.
Copyright (C) Sport Trades Ltd 2017 - 2020, John Keys and Jaskirat Rajasansir 2020 - 2024