Category: Allgemein

GSOC 2019 is drawing to a close and with it the first part of the HTTP API. To be completely honest this is not where I’d hoped to be at this stage, but we’ve set up a solid foundation for future work to build the rest of the API.

A Framework

I wrote about choosing a web framework a while back. The choice I arrived at, Iron, was in retrospect not the best choice. I chose it because I liked its middleware model, but it has only recently been renewed and most of its ecosystem hasn’t been updated in three years. This meant writing a lot of new middleware for the API. Additionally the only existing Iron testing framework was difficult to use and relied on fragile string manipulation to generate Request objects. I wrote a new testing framework called Anneal which uses hyper to generate Request objects and follows a builder pattern to simplify testing.

A Service

The HTTP Api is designed to operate as an independent service within a Qaul instance. An instance may disable the API, or not ever include it, and things should still work. Currently the mechanism by which services communicate is still a work in progress but a big part of the api was trying to pull as much of the boilerplate code used for validating incoming requests into the http-api service as possible.

Authentication is handled by the api, parsing of JSON:API requests (using the json-api crate I talked about in a previous post) is handled by the API, scoping of requests is handled by the api. The benefits of this model will become more apparent as we start to give services HTTP APIs.

A Plan

While my work on Qaul under GSOC is coming to a close I fully intend to complete my work on the HTTP API. I have a branch for adding unit tests to the api waiting on my user creation merge request, I have been experimenting with implementing an inter-service messaging system.

I added an http-api service, built login and logout endpoints, built a hot pluggable mount middleware for use in mounting services, implemented middleware for dealing with cookies and authenticating with cookies. I have written extensive error messages and documentation for all of these components and hopefully the foundation they have created will allow future development of the API to proceed with ease.

So long, and thanks for all the fish.

GSoC 2019 is coming to an end and for this reason it is unfortunately my last blog post on freifunk. For this reason, I would like to start by thanking the freifunk community, Google and, in particular, my mentors for the opportunity to participate in this rather special program. In this post I will present what has been done, what has not worked out and what is still to be done.

Code and APK

You can find the application code at my github repository:
https://github.com/kdebiec/GSoC_2019
And the application itself, ready to use here:
https://github.com/kdebiec/GSoC_2019/releases/tag/GSoC_2019_final
(Please note that you need the “Retroshare Service/Core” application in order for the application to work properly. – backend application that you can build from the Retroshare repository: https://github.com/RetroShare/RetroShare)

Overview

The aim of my project as I mentioned in my first post is to build a mobile application oriented around the chat, with a big focus on the modern look. The new logo, which is also part of this project, was intended to reflect new direction of this software.

Logo

Below you can see the new Retroshare logo, which is the main logo for this application, and in other projects its use will depend on the acceptance by users and the profile of the project.

• 

• 

• 

App

I have posted my proposed design in this post. Now it’s time to show off final design, but it is difficult to describe the appearance of the application and it makes little sense to paste several screens here. Because of this I recorded a walkthrough of the application. You can check it out here:

Illustrations of empty screens are provided by Icons8. Thanks for them!

Application was fully written in Dart using Flutter. This should ensure relative code readability, an optional ability to port the code to iOS, and relatively fast application performance.

Features

Functionalities that are now available in the application:

• add friend via certificate and share ours,
• create with avatar, change and delete our identities,
• see friends locations,
• create public room (lobby),
• send and receive message,
• add and remove contact
• search for chats, contacts, people,
• discover public rooms,
• see rooms participants,
• signin, signup;

Roadmap

There are still many interesting things to do in this project and for this reason I intend to continue my work. In particular, I would like to focus on these features:

• Make a good use of and improve Retroshare’s Json api event system. This will enable app to have message notifications and optimize its performance.
• Bundle backend and frontend into one app. Current system is confusing and leaves much to be desired. Who wants to manage the process themselves that is still in the background anyway?
• Explore possibility to add tor option as it is in Retroshare desktop app.
• Improve chat backend, especially much needed storage of history.
• After merging Retroshare’s short certs, add QR code as a default way to add friends. This will involve redesigning the way it works now.
• At this point, user still have to be aware of how Retroshare works under the hood. Future improvements have to be made so user will have to only operate with identities. For example, now, after adding friend we have to yet find his identity in search box to add him to contact and finally after those steps we can message him.
• The website is already a bit outdated and it could use a new look. As soon as the application is ready, I would like to refresh website.
• As you can see on video, rooms name are also ‘Error’, which of course is not the name of room. Due to lack of native support of 64int in Dart, lobbies ids can’t be loaded and so does names. Thanks to my mentor, Gio, solution is already in PR, and after merge, some minor changes have yet to be done on frontend side.

Conclusion

During this program I tried to build a very good application base that can be further developed, which I hope will make futher improvements easy. Unfortunately, during this period I was not able to meet all the milestones, specifically, the chat backend has not been improved. In spite of everything, I intend, according to my Roadmap, to improve the chat backend as well as add more functions to the app so that it can be considered as a production version.

I encourage everyone, especially current Retroshare users, to test the new app. I hope you like it.

Once again, I would like to thank my mentors for their help in recent months.

During this coding phase I added OpenWrt Makefiles to package conTest and the attenuator control software.
In addition some documentation was added, but most of the time I spend chasing the down the errors mentioned in the last blog post.

This error seems to be, that a certain attenuation gets repeated, while the config says something different.
After some long error hunting sessions RegMon[link] was added to the testbed to retrieve more information on the connection and hopefully
the error itself. RegMon allows to monitor the time consumption of ath9k wireless cards In there the error does not show up. While in
figure 1 the attenuation for signal 1 seems to be repeated, the RegMon diagramm (fig 2) does not show this behaviour. In figure 2 you can see
the time a wireless card spends sending (red) or receiving (blue) data, idling (yellowish) or taking care of interference (pink).
If the connection is dampened the needed time for sending/receiving data increases, as a more robust MCS, spatial stream and guard interval
combination is selected. Based on figure 1 I would expect the cards to have a higher busy time portion than shown in figure 2 at the problematic section.

Unfortunately an update broke the functionality of the regmon evaluation scripts, so I started to port it from R to Python3 to use it in future debugging sessions.
Figure 3 shows console output from the control program. The shown attention values are acquired from the digital attenuator shortly after the value was set
by the software. These values are as expected and show the correct behavior. The difference between the attenuation values on signal 0 and signal 1
originate from an added analogous attenuator.

Interestingly, after several tests with linear increasing/decreasing attenuation, the behaviour on figure 1 could not be observed.

I started to dig into the driver software of the Vaunix Labbrick attenuator, but did not find problematic code so far. But the first next step should be,
that the RegMon evaluation script is usable again. After that I will continue to look into the driver of the digital attenuator and in the worst case dive
into tcpdump, to see how it acquires the signal strength values.

The GSoC program is about to finish, and this will be my last GSoC-related blog post for Retroshare’s new web interface.

I will use this blog to provide an overview of how the app works, all my work done during this period, features, completed milestones, what couldn’t be completed, and future roadmap. I will also explain and document the code structure in the hopes that potential contributors will find it easy to get started.

• The source code is present here: https://github.com/RetroShare/RSNewWebUI
• And my fork over here: https://github.com/rottencandy/RSNewWebUI
• If you want to see the work I have done exclusively during GSoC, it is all in a single Pull Request: https://github.com/RetroShare/RSNewWebUI/pull/1

Overview

The purpose is very simple; A web-app that can be used to manage your Retroshare node, interact with friend nodes, and make use of Retroshare’s features. In other words, an alternative to the Qt-based interface of the client app.

This is made possible through the JSON API provided by Retroshare, which allows everyone to utilize the power of Retroshare’s technology to create their own services, interfaces, or even build apps on top of Retroshare.

The web interface itself works in a pretty straightforward manner, making use of modern browsers to act as a front end for the Retroshare platform and it’s services. Made using JavaScript, and the only external library being used is Mithril, which is a very fast and lightweight framework for building single page web applications.

Build process

If you look at the source code, you can see that it is built using qmake, the config file webui.pro executes build scripts in webui-src/make-src.

webui-src/
├── make-src/
│   ├── build.bat
│   ├── build.sh
│   ├── chat.css
│   ├── init.bat
│   ├── init.sh
│   ├── main.css
│   ├── readme.md
│   └── template.js

The build scripts in webui-src/make-src (most notably build.sh) iterate over all files from the source directory(webui-src/app), copying files into their respective destinations.

All JavaScript files are compiled into app.js and CSS files into app.css, these compiled files are put into the destination directory which is webui. The build scripts also copy all the static files, from webui-src/assets over to the destination directory, maintaining their directory hierarchy. Static files are the ones that do not require any modification in order to be used, like the HTML, font files, some CSS styles, and so on.

Another important aspect of the build process is how it compiles all the JavaScript files into a single file.
Since CSS is simply a set of rules without any structure, the output file can be built simply by appending all the source .css files together, JavaScript however doesn’t work that way:

You may have noticed another file in make-src called template.js. This file is used to create an entry point for the JavaScript files. It can be thought of as a kind of polyfill for require. What this essentially does is, take all the .js source files and store them using objects in such a way that they are isolated from each other, and then enable interaction between them through exporting objects.

To make a module’s components public, we have to refer them in the module.exports object, and we can use them inside other modules by importing them using the require() method.
The module.exports object is the only data that can be accessed outside of the module.

Structure

Now that we know how require() and module.exports work, we can look into how the source code functions:

app/
├── channels/
│   └── ...
├── chat/
│   └── ...
├── config/
│   └── ...
├── files/
│   └── ...
├── home.js
├── login.js
├── mail/
│   └── ...
├── main.js
├── mithril.js
├── network/
│   └── ...
├── people/
│   └── ...
├── rswebui.js
├── theme.css
└── widgets.js

The source files are all in webui-src/app. I have tried to implement a structure loosely based on the MVC design pattern. Aside from separating data and views as objects and components, it makes intuitive use of Mithril components and routing concepts.

Each folder contains the views and models for a single tab. All tabs have their own route resolver that takes in the route parameters and resolves them to return the correct views for rendering.

The entry point of all tabs happens in the resolver file, which also defines the layout of that tab. I will explain layout types shortly.

Note how the file names consist of their respective tab names too. This is not just for convention, our require polyfill does not yet have the concept of directory structure, any file present in any directory and may be imported by using only it’s name. This causes issues when accessing files having same names, which is why tab name supersedes the file name. It is important to have unique names for all files.

The main.js file contains mithril’s m.route, that defines the routing table and allows all navigation on the app. It detects whether the login keys have been verified and upon failing, reroutes to the login page using the onSuccess() callback.

The rswebui.js contains methods that act as the bridge between the web interface and the Retroshare client. Mainly, abstracting the API calls and managing async background tasks.

In a previous post, I mentioned that I did a lot of reading on UI and UX design, highlighting how it shaped the look and feel of the web interface.
After learning the importance of consistency when studying interaction design, I set out to make the interface more consistent.

In a nutshell, consistency refers to having uniformity in the UI, a form of repetition such that an action becomes predictable and intuitive to the user.

It can easily be achieved by having a predefined set of rules on how the UI should behave when interacted with, and the best way to do this is to make a set of reusable components. And since this is about the UI and visuals, it has more to do with CSS than JavaScript. Most of the layout rules are defined in theme.css.

The CSS class that houses all other widgets is the tab-page class. It’s the one containing all the elements under the navbar. All top level tab layouts use this and extend upon it. It can hold both full and half-width widgets, and position them according to the space taken by each.

The next is the sidebar class, which defines the sidebar on the left of some tabs, allowing to choose sub sections within the tab. It must be used when there are multiple sections but are logically grouped inside the same tab. Since this is a very commonly used widget, I have put this inside widgets.js, a file that contains a collection of the most used components, so that it is easily available everywhere. It takes in two parameters: the list of sections, and the base route link of the tab. Note that the section link must be the same as it’s name for it to be resolved properly:

m(widget.SideBar, {
    tabs: tabsList,
    baseRoute: '/route/',
});

The widget class is used as a preliminary frame for displaying small groups of input types together. For consistency, a directly interactable input must never be shown directly inside tab-page, but must be inside at least one widget frame. For additional uniformity, I have been using the <h3> followed by the <hr> tags as the immediately following elements to display and categorize a frame’s contents.

The progressbar widget is a combination of a <span> tag relatively placed inside a <div> tag using the block-inline display attribute. To create the progressbar in mithril, just use:

m('.progressbar', {
    style: {
        content: rate + '%'
    }
}, m('span.progress-status', {
    style: {
        width: rate + '%'
    }
}, rate));

And the CSS will handle the rest.

The tooltip widget can be used to display additional content when the mouse pointer is hovered over it. And can be easily created using:

m('.tooltip', [
    m('p', 'normally visible content'),
    m('.tooltiptext', 'content visible when hovered'),
]);

A modal or popup box can be used to display content which might be triggered by a user’s action, or can display information that requires immediate attention of the user. This is also present in the widgets.js file. It is made as a mithril component, so can be used normally with the m() selector. It also takes in other mithril components as attributes, allowing it to display any given html tag.

widgets.popupMessage([/* Array of components to render */]);

Creating custom input types making use of the <input> tag is incredibly easy in mithril, but may initially be confusing to programmers used to vanilla JavaScript for event handling.
Normally, to create a JS-controlled input field, you would do something like:

let text = document.getElementById('input').value;

But mithril components can be controlled very easily by making use of the onchange and oninput event handlers:

m('input[type=text]', {
    value: text,
    oninput: (e) => text = e.target.value,
});

And text gets automatically updated with any value the user enters. Use oninput to get value after all the text is entered, and onchange for more finer control, which is fired every time a button is pressed. This method can be used with any input type like text, number, radio, checkbox, etc. and can be made to run any function, allowing for huge flexibility and control.

Features

All the features and milestones that were successfully completed:

• Get your certificate, add new friends by copying in their certificates.
• View, manage your identities and get info about friend identities.
• View and manage all your friend nodes and each of their locations, and basic info about them.
• Get info about your upload/download files and manage them, add new downloads through links.
• Check all your mails.
• View info about subscribed chat lobbies and publicly available lobbies.
• Change various configuration options of your Retroshare node such as network limits, file locations, default behavior, and such.

Roadmap

Retroshare has a huge amount of features, and unfortunately this period wasn’t enough for me to cover all of them into the web interface. I plan on implementing the incomplete tabs and then extending the app with new functionality:

• Turtle search:
  As my mentor Cyril told me, this feature is very important since it makes it very easy to find and download new files, and is one of the features making use of stream data from the API.
  Getting stream data has been a problem due to CORS implemented on browsers, which is the reason this feature couldn’t be finished. I am constantly looking for a viable solution and will immediately finish implementing this when I find one.
• Sending mails:
  The web interface can only read mails for now, and it would be very nice to be able to send mails too.
• Forums:
  I have already started work on the forums tab, and will finish it soon. This will allow users to interact with and manage forums entirely from inside the web interface.
• Channels:
  Similarly, I am also working on channels. Another nice feature to have on the Web UI.
• Build Process:
  As shown above, the current build process is very barebones, and the require polyfill has no concept of directory hierarchy. This will eventually cause issues as the app grows. We need to upgrade the build tools, or find a new one. I think the most important point to keep in mind if choosing to go with a new one, is that the user should not have to install any additional dependencies.

That’s about it. I encourage everyone to try out the app, it is very easy to install the web interface. There are even simple installation instructions on the source page! Feel free to get in touch if you have any suggestions or queries. You can generally find me lurking in the Developer forums in Retroshare.

Many thanks to Google, and the amazing Freifunk community, especially my mentors, for giving me this opportunity. This has been a wonderful time for me, I learned a lot of new things that would help me contribute more towards free and open software.

Hi all!

On this final evaluation we focus the work on give to the automatic API wrapper generator support for openapi-swagger specification.

As Wikipedia said:

The OpenAPI Specification, is a specification for machine-readable interface files for describing, producing, consuming, and visualizing RESTful web services.

In addition, there are lots of tools that generate the front and backend code automatically from an OpenAPI specification. So, theoretically, an API interfacedefined with the OpenAPI specification, can easily be generated in an amount of languages using the OpenAPI Generator Project .

So with my mentor decided that this is more useful and priority for RetroShare project, because anybody could generate their own wrapper in a lots of languages in an automated way from the last commit of RetroShare master branch without write any code. 

So the workflow should be:

1. Compile the RetroShare desired commit to generate the Doxygen documentation.
2. Use the jsonapiwrapper-generator-openapi.py script that generates the OpenAPI YAML specification for RetroShare API.
3. Generate the frontend code using  the OpenAPI Generator Project on to the desired language. For example, you can check the result for python on this example.

Following this steps, easily you can have an updated RetroShare API wrapper without coding anything!.

Some interesting tip about auto generated code is that the documentation are generated automatically with working examples.

Workaround

After we decided to give the support to OpenAPI specification we start to adapt the previous script https://gitlab.com/jpascualsana/retroshare-api-wrapper-generator/blob/master/jsonapiwrapper-generator.py to support the new specification. 

Trying to use the generated YAML against the generator validator shows us a lotsof warnings and crashes of our script. So we need to solve lots of bugs that we didn’t know that exists, bugs related of RsClasses generation, types etc… https://github.com/RetroShare/RetroShare/pull/1614

Also we deeply studied how to translate the RS API on to the OpenAPI specification, studying how to translate the types, types of functions, the authentication etc… 

Then, how to integrate the documentation on to the YAML to see the result on the generated code, following the specification. 

Finally we get in trouble with the asynchronous functions, that are not fully supported for the python generator and we are already looking for a solution.

So once we had the generated API wrapper we start to test it:

• Tests for unauthenticated methods

• Tests for login

• Tests for authenticated end points.

• Tests for models: when the model objects are created using the OpenAPI Generator, you can pass an entire object as dictionary or instantiate it with the provided Class created by the generator. For example, on the link above we can see how to instantiate the object instead of pass it as dictionary.

groupMetada = RsGroupMetaData(m_group_name="Just another test channel2", m_group_flags=4, m_sign_flags=520)
channel = RsGxsChannelGroup(m_meta=groupMetada, m_description="Test it!")
req_rs_gxs_channels_create_channel = {"channel": channel}

• Tests for publish post with file upload: look at the commit , first the file should be on a shared directory to be shared properly.

Conclusion

So on this GSoC we started thinking that we  are going to code a way to import public datasets on the RetroShare network .

To do that we understood the necessity of create an automatically API wrapper, to do not code by hand each end-point of the API. 

Finally we found the necessity to provide a way to generate wrappers on any language needed on an automatic way, finding an already created solution on the OpenAPI Generator Project. 

Now is easier for developers to write applications over RetroShare network using the RetroShare JSON API, unifying the wrappers generation. 

On the back we let the project to import pulic datasets that I would like to finish on the following months, after GSoC, implementing the OpenAPI wrapper to the “import public datasets” repository. 

Repositories

• public-datasets-import
• retroshare-api-wrapper-generator
• openapi-python-retroshare-api-wrapper

Freifunk posts

1. Arrival
2. First evaluation
3. Second evaluation
4. Third evaluation

GSoC report

https://gitlab.com/jpascualsana/gsoc-2019-report

During this Google Summer of Code, I built visn, a testing framework designed specifically to integration-test Rust projects that rely on eventual consistency.

Originally, I imagined that a full network simulator was needed to thoroughly exercise the qaul.net API, as I discussed in my initial blog post. While sketching out designs for the libqaul API, as seen in this pull request (which influenced the actual service API), I discovered that it made much more sense to simply simulate the events coming into an individual instance. This is simpler to design, easier for developers to use, and less computationally complex.

That insight lead to the idea of a generic, easy-to-extend testing framework which could be used to test qaul at multiple levels at once. This framework, discussed in more detail in my second blog post, essential boils down to a mechanism for writing tests in an easy-to-read format which is transformed into calls into the system under test.

While working on this idea, I was also involved in several pull requests which added various capabilities to the libqaul API, including adding the Message struct and designing security and consistency validation steps for messages.

Eventually, my first pull request for visn itself came along. I got a lot of useful feedback and decided that the easiest way to make sure I was writing the right kind of test framework was to actually use it, by implementing features in libqaul and testing them at the same time as I implemented features in visn. This lead to my third blog post, on the value of this approach to what I call “conceptual” testing and my first PR that actually implemented and tested some libqaul features, along with the PR that fixed the issues I mentioned in that blog post.

Around this time, the project moved to a GitLab instance, so my PRs were somewhat split. In addition to a first stab at a contacts book API, which I later revised, I added the most recent visn feature, and in my opinion, one of the most useful: permutation testing.

In essence, rather than simply testing a single order of events, visn now allows testing all possible orderings of a set of events. This is, of course, and O(n!) algorithm, since there are n! orderings of n events, but I was able to do it pretty efficiently with the use of what eventually became the permute crate.

permute uses a data structure, called an ArbitraryTandemControlIterator, that stores both a reference to an array (a “slice”) and an iterator over indices to that array and transparently iterates over references to elements in that array. This way, copying of the array’s elements is kept to an absolute minimum, reducing both computational and memory footprint.

For example, reversing the elements of a vector using an ATCI:

let data = vec![
    String::from("red"),
    String::from("orange"),
    String::from("yellow"),
    String::from("green"),
    String::from("blue"),
    String::from("indigo"),
    String::from("violet"),
];

let control = vec![6, 5, 4, 3, 2, 1];

let atci = ArbitraryTandemControlIterator::new(&data, control.clone().into_iter());

for (atci_val, rev_val) in atci.zip(data.iter().rev()) {
       // Critically, these are both &std::string::String. No copying occurred.
       assert_eq!(atci_val, rev_val);
}

Of course, control could just as easily be vec![6, 5, 6, 4, 4, 3, 2, 1] or any other combination of valid indices.

Combined with an implementation of Heap’s algorithm for permuting lists, permute can provide an efficient method of generating all possible orderings in a deterministic way.

With that done, I was able to add a sample all-orders test for some aspects of the libqaul API, and tidy up the internal testing crate so that others can use this work. Once these two merge requests are reviewed and accepted, the visn testing framework and the service_sim crate in the qaul.net tree will be fully able to support ongoing libqual development.

Google Summer of Code with Freifunk has been a wonderful experience during which I’ve worked with cool technologies, amazing peers, and extremely helpful mentors. Overall, I think it’s made me a much better technologist and programmer, and I’d like to wholeheartedly thank Google, Freifunk, and my mentors for the amazing opportunity.

While working on the visn eventual consistency testing framework for the qaul.net project, I’ve run into an excellent example of one of the most important reasons to test software, in some ways more important than the discovery of regressions, design defects, or other functional issues. Specifically, the ability to determine problems in the conceptual model around which the software is built.

Conceptual Testing

Unit, integration, and acceptance tests are well known for their value in detecting regressions, ensuring that functions, classes, and other units are written in a self-contained and composable style, and ensuring that design goals are met throughout the lifecycle of the project.

In statically typed languages like Rust, however, it can often be tempting to eschew the fine-grained level of unit testing used in dynamic languages, since the compiler checks many of the constraints unit tests are designed to impose. Rust, for example, permits encoding a lot of detail about the presence, or absence, of values with the type system.

In qaul.net’s libqaul, we provide a model for metadata about a user in the UserData struct (from libqaul/src/users/mod.rs):

/// A public representation of user information
///
/// This struct is used for both the local user (identified
/// by `UserAuth`) as well as remote users from the contacts book.
#[derive(Default, Debug, PartialEq, Clone)]
pub struct UserData {
    /// A human readable display-name (like @foobar)
    pub display_name: Option<String>,
    /// A human's preferred call-sign ("Friends call me foo")
    pub real_name: Option<String>,
    /// A key-value list of things the user deems interesting
    /// about themselves. This could be stuff like "gender",
    /// "preferred languages" or whatever.
    pub bio: BTreeMap<String, String>,
    /// The set of services this user runs (should never be empty!)
    pub services: BTreeSet<String>,
    /// A users profile picture (some people like selfies)
    pub avatar: Option<Vec<u8>>,
}

This struct provides Optional types for every field, except for those fields which can contain nothing (like the BTreeMap or BTreeSet), since by design, a user may not have set any of these fields. This works really well for user storage, which was the original purpose of the data structure, but does not work well for user information transmission, as I found out.

Conceptual Problems in the User Model

Initially, libqaul was designed to use the UserData struct for all user data needs, including transmission. Some of the UserData related API surface was the first that I implemented during the initial deployment of visn, and therefore the first API surface to be tested. During that process, I mapped the function Qaul::user_update() to a visn synthetic event, UserUpdate, which carried a UserData to be passed to user_update().

While writing these tests, I encountered a problem: what happens in the case that a user wants to clear a field in their UserData? Do they issue a UserData in which there is an Option::None value in that field (like a null), which is interpreted to mean that the field should be cleared?

This made the user_update() function very easy to implement, since it could simply assign the newly received UserData as the new canonical UserData for that user. That, however, leads to a problem when it comes to data transmission over the actual network. When, for example, a user has set a profile photo or a lot biography fields, the UserData could be pretty large, and retransmitting that on every subsequent update is not very practical.

The act of writing these tests, which were primarily designed to prevent regressions, lead me to implement a delta-based UserData update scheme, wherein the UserData is updated incrementally with small changes. This provides other advantages, too, such as allowing more orderings of those events’ arrivals to result in a valid state for the UserData.

Conceptual Problems in visn

In addition to uncovering problems in the design of libqaul, this process helped me refine my ideas for the visn testing framework. Initially, visn assumed that all operations modelled by synthetic events were infallible, or at least that failure to perform an action should lead to test failure. In fact, a critical component of eventually consistent systems is their ability to reject invalid states, in order to remain robust in the face of serious network problems or malicious input.

Originally, the resolve function took the state of the system under test and an event, and returned the new state (Fn(Event, System) -> System).

To address this problem, visn‘s type model became even more complex, incorporating a separate return type rather than requiring that the function which resolves events always return a successfully transformed state (Fn(Event, System) -> Return), and the infallible variant now simply sets Return to System.

In addition, rather than taking a singular System argument when resolving events, visn now takes a function returning a System, laying the groundwork for supporting multiple permutations of the ordering of queued events.

Conclusion

Testing is important for both compiled and dynamic languages to prevent defects and enforce good factorization, but the benefits to compiled languages can, like many design processes, be moved “left”, into the pre-execution step. As seen here, the simple act of writing tests often leads to conflict with the type system and compiler that can reveal conceptual and design defects in the system being tested.

The new qaul.net HTTP API speaks JSON, as do increasingly many things. It allows you to express complex types, it maps well to most programmers’ mental models, it’s self describing, and there’s a decent library for it in every language under the sun. In the Rust world JSON is primarily dealt with using the serde_json crate (https://crates.io/crates/serde_json) which allows the programmer to easily map strictly typed structures into JSON and back. Today we’re going to be talking about the difficulties we encountered building a type for JSON:API’s relationship data field (https://jsonapi.org/format/#document-resource-object-linkage).

The data field of a relationship can be any of the following things:

• non-existant
• null
• a single object
• []
• an array of objects

Each of these options semantically represents a distinct thing and so we should be able to tell them apart. We will use the following enumto represent our value:

#[derive(Serialize)]
#[serde(untagged)]
enum OptionalVec<T> {
  NotPresent,
  One(Option<T>),
  Many(Vec<T>),
}

Non-existant relationships will be represented by OptionalVec::NotPresent, to-one relationships (empty or otherwise) will be represented by OptionalVec::One, and to-many relationships will be represented by OptionalVec::Many.

Serializing

To allow our enum to serialize properly we just need to implement a method to tell if it’s supposed to be present or not:

impl<T> OptionalVec<T> {
  pub fn is_not_present(&self) -> bool {
    match self {
      OptionalVec::NotPresent => true,
      _ => false,
    }
  }
}

Now when we wish to use our enum we simply need to put the #[serde(skip_serializing_if = "OptionalVec::is_not_present")] attribute before the field.

Deserializing

To cover the OptionalVec::NotPresent case we will need to implement std::default::Default for OptionalVec as follows:

impl<T> Default for OptionalVec<T> {
  fn default() -> Self {
    OptionalVec::NotPresent
  }
}

Now whenever we use OptionalVec we need to add the #[serde(default)] attribute to the field. This tells serde to fill the field with a default value if the key isn’t present.

For the other options, we need to implement a custom deserializer. The technically proper way to do this is to build a Visitor, but we’re going to take the simpler route and deserialize it to serde_json::Value first. Our deserializer is as follows:

impl<'de, T> Deserialize<'de> for OptionalVec<T>
where T: DeserializeOwned {
  fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
  where D: Deserializer<'de> {
    let v = Value::deserialize(deserializer)?;
    match serde_json::from_value::<Option<T>>(v.clone()) {
      Ok(one) => Ok(OptionalVec::One(one)),
      Err(_) => match serde_json::from_value(v) {
        Ok(many) => Ok(OptionalVec::Many(many)),
        Err(_) => Err(D::Error::custom("Neither one nor many")),
      },
    }
  }
}

And that’s it! Effectively we try first to deserialize the singular case and if that fails we try to deserialize the multiple case. The first case will catch null as Option<T> will deserialize null as None.

Conclusion

This is just one of the many challenges we faced writing a framework for JSON:API parsing in Rust. The contents of this article in their proper context can be found here: https://github.com/qaul/json-api/blob/master/src/optional_vec.rs