Hoist is a web application development toolkit developed by Extremely Heavy.
Hoist is designed as a "full stack" UI development framework, meaning that it has both server and client components that work together to provide an integrated set of tools and utilities for quickly constructing sophisticated front-end interfaces - or entire applications - with a strong focus on building for the enterprise.
The core technologies used are Java and its more dynamic cousin Groovy - via the mature Grails framework - and JavaScript - via React and associated libraries.
This repository is hoist-core, which is the server-side implementation of Hoist. It is designed to be used with hoist-react, our latest front-end toolkit. Hoist's original front-end implementation. See that repository for detailed information on Hoist's client-side features and conventions.
This readme is intended to provide a general orientation to the features provided by Hoist, notes on specific features and implementation details of particular interest, and direct links into the most relevant and commonly used source code classes and routines. We have aimed to make the code itself as clear, readable, and well-commented as possible, and we are working to ensure a consistent level of in-code documentation - especially at the class level and for essential public methods.
When calling out particular source code files in the tables below, the following icons are used as the links to the containing repository: 🏗 for this repo and ⚛ for hoist-react.
This document does not provide an overview or introduction to Grails, Java, or other core technologies in general, and assumes a general familiarity with many concepts related to enterprise web application development.
Hoist grew out of our ongoing practice developing applications for enterprise clients - primarily in finance - that required multiple interrelated yet distinct applications that were:
While there are clearly many application development libraries and frameworks, we required a toolkit that could pre-select a set of libraries and bring together higher-level services such as:
With Hoist, these features work together and build on each other. A simple utility method to make an Ajax request to the server can automatically decode a JSON response, save a tracking record of who made the call and how long it took, and clearly alert the user and/or app administrators if there was an error. A customized grid component can offer a full-featured UI for column selection, support filtering column choices based on user roles, and persist the user's choice of visible columns and sorting options as a preference that is maintained across browsing sessions and workstations.
The primary focus of Hoist is on building user interfaces that sit in front of and display data from a variety of back-end sources and services already deployed within an enterprise. As such, much of the developer's interaction with Hoist is in the form of client-side JavaScript development.
Given this emphasis on the front-end, the role of the Grails server provided by hoist-core can be limited almost exclusively to providing the built-in infrastructure expected by the client-side toolkits - serving up and storing configuration, preference, tracking, and other related data. These tasks require a server capable of persisting data to a database - Grails supports a wide variety - and of securely processing requests and serializing data.
When it comes to fetching business data specific to an application, a Hoist JS app can talk to independent back-end systems directly via CORS or proxy arrangements. Indeed once an initial project setup is complete, a developer might never touch Java / Groovy code.
That said, Grails does provide a feature rich server layer with the full power of the Java ecosystem available for use. Roles played by the Hoist server commonly include:
A Hoist app is structured primarily as a Grails 6.x application, with a file and directory layout that follows the Grails conventions. The Grails project offers extensive and well maintained documentation on the framework's features and configuration. This library - hoist-core - is packaged as a Grails plugin, and should be specified as such within the Grails app's build.gradle file, e.g.:
dependencies {
... // standard Grails dependencies / plugins / app-specific libraries
compile "io.xh:hoist-core:$hoistCoreVersion"`
}This will add server-side support for Hoist's core back-end services, including a set of endpoints expected by the client-side toolkit implementations. Versioned and snapshot builds of Hoist are pushed to XH's maven repository, which must also be configured within build.gradle.
Grails applications are built via Gradle, a highly flexible and popular build tool. The result is a single WAR file which can be deployed via Apache Tomcat.
All client-side code is commonly maintained within the same repository, but within its own dedicated folder tree. For hoist-react applications, JS apps are built independently by Webpack and deployed via nginx. We recommend and provide standardized Docker containers to ship both sides of the application and tie them together in an integrated whole, with very minimal infrastructure requirements or dependencies. See the Hoist React readme for a detailed outline on the build and deploy process for React-based applications, including information on building the Grails server-side components outlined here.
While Hoist Core and its associated client-side libraries provide a good deal of functionality on their own, we realize that enterprise clients will have configurations, authentication requirements, dependencies, data source definitions, and other reusable code constructs that are unique to their environment.
To support these needs while still encouraging maximum consistency across multiple applications, we can assist in creating and maintaining a custom plugin layer between Hoist and business applications.
Hoist is currently developed exclusively by Extremely Heavy and intended for use by XH and our client partners to develop enterprise web applications with XH's guidance and direction. That said, we have released the toolkit under the permissive and open Apache 2.0 license. This allows other developers, regardless of whether they are current XH clients or not, to checkout, use, modify, and otherwise explore Hoist and its source code. See this project's license file for the full license.
We have selected an open source license as part of our ongoing commitment to openness, transparency, and ease-of-use, and to clarify and emphasize the suitability of Hoist for use within a wide variety of enterprise software projects. Note, however, that we cannot at this time commit to any particular support or contribution model outside of our consulting work. But if you are interested in Hoist and/or think it might be helpful for a project, please don't hesitate to contact us!
While this document does not aim to provide a definitive or complete guide to the components of the Hoist framework or its usage, several key features are called out with additional details below.
| Class/File | Note | Link |
|---|---|---|
BaseAuthenticationService.groovy |
App must implement to define auth scheme | 🏗 |
BaseUserService.groovy |
App must implement to generate users | 🏗 |
HoistUser.groovy |
Trait/interface for core user data | 🏗 |
IdentityService.groovy |
Server-side source of current user info | 🏗 |
IdentityService.ts |
Hoist-React source of current user info | ⚛️ |
👫 As organizations and applications will have a wide variety of requirements for
authenticating and authorizing users, Hoist has a deliberately minimal interface in this regard. A
primary requirement for applications is that they implement a Grails Service named
AuthenticationService that extends Hoist's BaseAuthenticationService and implement its
completeAuthentication() method.
Implementations of this method must lookup and/or create a User class which implements the
HoistUser trait/interface. This specifies the core information Hoist expects to access for any
logged in user. Applications can choose to enhance their own user class with any additional details,
managed via the app or sourced from systems such as Active Directory / LDAP.
While not included in Hoist directly, NTLM / SSO can be supported via integration with the Jespa library, commonly done via a custom plugin.
Once authentication is complete, IdentityService is the primary server-side Service for getting a
reference to the current user. Hoist's client side code calls a dedicated endpoint to verify and
fetch core user info, making it easily available to the JS app via a corresponding JS service.
| Class/File | Note | Link |
|---|---|---|
BaseRoleService.groovy |
App must implement to assign roles to users | 🏗 |
Access.groovy |
Annotation for endpoint security | 🏗 |
🔒 Structure is provided for application "roles", for use in defining access to various parts of the
application. At their core, roles are simply strings -- it is up to the application to determine what
meaning they will have in the context of the app. Note that Hoist will expect (and auto-create) certain
roles (e.g. "HOIST_ADMIN", "HOIST_ADMIN_READER", "HOIST_IMPERSONATOR", and "HOIST_ROLE_MANAGER")
that protect built-in hoist functionality and should be assigned as needed to administrators of
the app.
Roles are associated with users via an application defined RoleService. Hoist provides an
out-of-the-box solution for this -- DefaultRoleService -- that will store Role definitions and
assignments in the database, can be integrated with other systems such as Active Directory, and
has a built-in management UI in Hoist React. DefaultRoleService may be customized, but for
applications that require a fully custom solution, BaseRoleService establishes the minimal required
API contract and may be subclassed directly.
Server-side endpoints (Controllers) can be restricted to users with a given role or roles via the
@Access annotation, e.g. a controller that should be accessible only to users with an "EDITOR"
role could be decorated as such:
@Access(['EDITOR'])
class ReportController extends BaseController {
def saveReport(params) { ... }
}The @AccessAll annotation allows any user access. A controller endpoint must be decorated with one
or the other of these annotations or an exception will be thrown.
😎 Administrators have access to an impersonation mode where they can "act as" another
user in the context of the application. This process is managed by IdentityService, which exposes
several public methods for entering and exiting impersonation mode. When active,
IdentityService.getUser() will return the user being impersonated, while
IdentityService.getAuthUser() will return the actual admin.
The client toolkits provide built-in UIs for administrators to enter and exit impersonation mode. Services such as activity tracking are aware of impersonation and will log activity done while impersonation is active with both the impersonated and real user.
| Class/File | Note | Link |
|---|---|---|
AppConfig.groovy |
Domain object for config entries | 🏗 |
ConfigService.groovy |
Server-side source for configs | 🏗 |
ConfigService.ts |
Hoist-React source for configs | ⚛️ |
🔧 The ability to store simple typed configuration values
(string|int|long|double|bool|json|pwd) and manage / adjust them in a running application has
proven to be an extremely useful core feature. AppConfig entries are stored in the UI server's
database and referenced via a simple string name.
Configs can also be made available to client applications (or not) via a dedicated flag, where they can be referenced by JS code. The built-in Admin console provides a full UI for reviewing, updating, and organizing these entries.
🙈 A special pwd type allows passwords and other sensitive info to be stored in an
encrypted form for use on the server, avoiding the need to save these common types of configuration
to the database in plaintext. Note however that any developer can deliberately print the output of
an encrypted config by logging the (unencrypted) output of configService.getPwd().
Hoist requires certain configuration entries to be defined and present for the application to
initialize. Apps themselves might also have hard dependencies on configs. To help ensure these
entries are in place and to aid in the spinning up of a new app with an empty database, the
ConfigService.ensureRequiredConfigsCreated() method is available to verify and auto-create
required configs. See Bootstrap.groovy in hoist-core for configs required at the Hoist level.
| Class/File | Note | Link |
|---|---|---|
Preference.groovy |
Domain object for preference definition | 🏗 |
UserPreference.groovy |
Domain object for user-specific value | 🏗 |
PrefService.groovy |
Server-side pref management | 🏗 |
PrefService.ts |
Hoist-React pref management | ⚛️ |
⭐ Preferences provide a lightweight way to persist user-specific options and settings. Similar
to AppConfigs, preferences offer several predefined data types
(string|int|long|double|boolean|json) and are referenced by a string name property. Preferences
are assigned a default value that is returned if a user does not yet have a specific value set. When
a user preference is assigned via PrefService.setPreference() (or one of the typed setters) a
UserPreference object is created and saved. Both objects can be managed via the built-in Admin
console.
Preferences may generally be accessed and used on the server and client, although they are primarily
a client-side tool. Preferences with the local flag set to true, however, have their user-specific
values stored on the client (in local storage) and are not accessible on the server. This flag is
designed for preferences that store things like layout or sizing information that are most
appropriate to save in the context of a particular device or workstation.
Applications are encouraged to provide end-users with controls to reset their preferences should
they wish to restore their profile to a default state. Server and client APIs exist to do such a
thing - see PrefService.clearPreferences().
As with configs (above) the PrefService.ensureRequiredPrefsCreated() method is available to verify
and auto-create required preferences. See Bootstrap.groovy in hoist-core for prefs required at the
Hoist level.
| Class/File | Note | Link |
|---|---|---|
TrackLog.groovy |
Domain object for track entries | 🏗 |
ClientError.groovy |
Domain object for error reports | 🏗 |
Feedback.groovy |
Domain object for user feedback | 🏗 |
TrackService.groovy |
Server-side API to log activity | 🏗 |
TrackService.ts |
Hoist-React API to log activity | ⚛️ |
ExceptionHandler.ts |
Hoist-React API to track errors | ⚛️ |
👀 Knowing which users are visiting an app and tracking specific actions of interest is another common need for apps. Hoist includes an API for easily tracking activity for the current user, and the Admin console provides a UI for searching and reviewing activity. Hoist services track some activities automatically (e.g. impersonation), but it is primarily up to app developers to determine which activities are of interest for tracking.
In its simplest form, a tracking record is a string - e.g. "Viewed chart". Tracks can also be given
a category for easier organization in the UI and a JSON map of data for additional details (i.e. to
note query parameters. The TrackLog object stores this record along with a timestamp, the current
user, and browser/device info.
On the client-side, a track() method is added to the Promise prototype to provide convenient
tracking for asynchronous requests - e.g. tracking a particular API call. This method provides
built-in timing of the call and saves as 'TrackLog.elapsed'.
💥 The ClientError object provides a special variation on tracking to handle exception reports
posted by the client applications. See ExceptionHandler.ts for the hoist-react entry point to this
service. Note that the ClientErrorService on the server fires an xhClientErrorReceived event,
which is listened to be the related ClientErrorEmailService to automatically send error reports to
the configured xhEmailSupport email address. Custom services can also listen to these events to
e.g. send other notifications via instant message, or raise an issue in a ticketing system.
💬 A simple model is also included for collecting and storing feedback (in the form of
simple messages) submitted by end-users directly from the application. A FeedbackService fires a
similar event and is listened to by a built-in service that will email out report notifications.
| Class/File | Note | Link |
|---|---|---|
EmailService.groovy |
Managed service for sending email | 🏗 |
📫 Hoist core provides EmailService to send mail from the server. This relies on the
Grails mail plugin which must be configured with a
suitable SMTP server within an app's application.groovy configuration file.
Several default AppConfigs are available to default the sender and provide filtering and override options for email, especially useful in dev/test scenarios where careful control of what emails are sent to which users is required.
| Class/File | Note | Link |
|---|---|---|
Monitor.groovy |
Domain object for monitor definitions | 🏗 |
MonitorResult.groovy |
In-memory object for monitor outcomes | 🏗 |
MonitoringService.groovy |
Service that coordinates monitor runs | 🏗 |
👍👎 Hoist provides an API and services for runtime monitoring of the application, with a deliberate focus on running application-specific checks that relate to the business logic and data sources specific to the app (as opposed to e.g. system or OS level monitoring of metrics like CPU or memory usage).
To use monitoring, applications must implement a MonitorDefinitionService (i.e. a standard Grails
service with that name) that implements one or more monitoring methods of the form:
// Method signature
def customCheckName(MonitorResult result) {
// Run any custom business logic here, setting properties on MonitorResult to record outcomes
}
// For example...
def activeTradeCount(MonitorResult result) {
def trades = tradeReportingService.getActiveTrades()
result.metric = trades.size()
}
def inventorAgeMs(MonitorResult result) {
def rptDate = inventoryWatcher.getReport().asOfDate
result.metric = System.currentTimeMillis() - rptDate.time
result.message = "Latest inventory update: ${asOfDate}"
}The names of these methods should match the code property of Monitor objects created and managed
via the Admin console. These definition objects hold data-driven parameters to determine how monitor
results are evaluated. The MonitorResult.status property is the outcome of a given monitor. While
the app's MonitorDefinitionService can set this status directly on results within its methods, a
more flexible and common pattern is to have the service set a metric instead - e.g. the number of
rows returned by a query, or the age in seconds of a particular result set.
This metric can then be evaluated against data-driven parameters on the Monitor object to
determine the status dynamically, allowing for runtime adjustments and tuning of the checks. The
Hoist monitor runner will time all checks (and enforce a timeout) and catch any exceptions that
might get thrown (marking the check as having failed and noting the exception on the result).
Monitor results can be viewed via the Admin console. The xhMonitorConfig and
xhMonitorEmailRecipients configs control option for email-based alerting on monitor failures,
including support for debouncing alerts. MonitoringService fires a server-side
xhMonitorStatusReport event that can be picked up by other custom services for additional
notifications.
🔮 Note an XH project is underway to provide a more general and cross-application implementation of this monitoring API for both Hoist and non-Hoist based applications.
Hot reloading is supported for both Java11 and Java17 using the java hotswap agent. Please see http://hotswapagent.org/ and https://github.com/HotswapProjects/HotswapAgent for details on how to setup.
Please note that the following is needed for running in this mode. See the toolbox application for an example of this setup.
groovyReset.jar in your project or dependencies-XX:HotswapAgent=fatjar, -XX:+AllowEnhancedClassRedefinition,
and -javaagent:absolute-path-to-groovyReset.jarNote that hot swapping is not currently enabled for grails plugins being developed in wrapper projects.
🚧 The sections below are planned near-term additions to this readme, covering additional features and services provided by the framework.
🙏 Thanks for your patience, and for reading this far!
📫☎️🌎 info@xh.io | https://xh.io
Copyright © 2023 Extremely Heavy Industries Inc.