Libervia progress note 2021-W38

goffi 27/09/2021, 06:54 jabber-xmpp-en SàT Libervia project libre Libervia progress SàT progress

Hello,

it's time for a new progress note. The work is currently focused on ActivityPub Gateway, and progress has been done on pubsub cache search and the base component.

Pubsub Cache Full-Text Search

Next to the pubsub cache implementation, it was necessary to have a good way to search among items.

So far, Libervia was doing pubsub search using pubsub service's capabilities, and notably the XEP-0431(Full Text Search in MAM) implementation. This is working well (it's what is currently used on this very blog when you do use the search box), but has some pitfalls: the pubsub service must implement this XEP (and as far as I know, Libervia Pubsub is the only one which does it at the moment), the search can be done in a single node at a time only, each search request imply a new XMPP request to the pubsub service, and pubsub items must be in plain text (which is currently always the case, but pubsub end-to-end encryption is planned as second part of the granted NLNet project on which I'm working).

In regard to that, a local search is necessary. SQLAlchemy doesn't really have Full-Text Search (or FTS) support for SQLite out of the box, but it allows to use any SQL directly, thus I could use the really nice FTS engine available within it (FTS5). This is an extension, but in practice it is already installed most of the time (it is part of the SQLite amalgamation).

Thanks to the JSON support in SQLite, it is also possible to filter search requests on parsed data. That's really useful for features like blogs where you often want to do that (e.g. filtering on tags).

The cache search can be operated on all data in cache, that means that you can do search on items coming from multiple nodes and even multiple services. That opens the door to features like hashtags or blog suggestions.

Last but not least, search requests can be ordered by any parsed field. In other terms it will be possible to order a blog by declared publication date — which may be important if you want to import a blog —, or events by location.

To have an idea of the possibilities, you can check the documentation of the CLI search command.

Base ActivityPub Component

Once the preparatory steps have been done, the ActivityPub component itself could be started. In short, for people not used to XMPP, a "component" is a kind of generic plugin to server. You declare it in your server configuration, choose a JID and a "shared secret" (a password), run it with those parameters, and voilà.

For the AP gateway, Libervia runs the component. There is documentation to explain how to launch it, don't worry it's simple.

As I've got questions about this, here is a small schema giving an overview on how the whole thing is working:

global overview of Libervia ActivityPub Gateway

I hope that it makes the whole thing more clear, otherwise don't hesitate to ask me for clarification.

As you can see, the gateway includes an HTTP server to communicate with AP software, but in many cases there will already be an HTTP server (website, XMPP web client, etc.). In this case, you'll have to redirect /.well-known/webfinger and /_ap requests to the gateway server.

For the development, I'm using Prosody as reference XMPP server implementation, and Mastodon as reference ActivityPub server implementation. I've set a local Mastodon installation, and I've chosen to use Docker for that, as it makes things easy to have a reproducible environment and to save and restore a specific state. It was not as trivial as I would expect to find the right configuration to use, I've found outdated tutorials, but I could manage to run the thing relatively easily.

Because we work with HTTPS, I've made a custom docker image with locale certification authority, so Mastodon could validate my gateway HTTP server certificate. I'm already doing that for docker image used for end-to-end tests of Libervia, nothing difficult. Surprisingly though, Mastodon could not resolve my instance, when HTTPie running from the same container could do it flawlessly. I've quickly realised that Mastodon was not respecting hosts declared in /etc/hosts (and added via extra_hosts in Compose file) and found a relevant bug report on Mastodon tracker. That was annoying, and I had to find a way to work around that. I've done it by running a local DNS Server, and Twisted offers a nice built-in one. Twisted DNS can easily use /etc/hosts to direct my local domains to my local IP, it's just a one liner such as twistd3 -n dns --hosts-file=/etc/hosts -r.

After that the domain was resolving, but to my surprise, Mastodon was still not able to communicate with my gateway, and even more bizarre my server was receiving no request at all. After a quick round of tcpdump/wireshark, I saw that indeed nothing was sent to my server.

Thanks to the Libre nature of Mastodon, I could resolve this by reading the source code, the Mastodon::HostValidationError
led me to a section that made the whole picture clear: my server is on a local IP and Mastodon by default refuses to reach it (to avoid the confused deputy attack). With the ALLOWED_PRIVATE_ADDRESSES setting I could finally make Mastodon communicate with my server.

The How to implement a basic ActivityPub server tutorial made by Eugen Rochko (Mastodon original developer) is a nice article to start an ActivityPub implementation, it has been useful to build the base component (despite being a bit outdated, notably regarding signature).

I have to rant a bit, though, as the ActivityPub specification are not available in EPUB or PDF, making it difficult to read on an e-book reader. I could overcome that thanks to pandoc (git clone https://github.com/w3c/activitypub.git then pandoc index.html --pdf-engine=xelatex -o activitypub.pdf), it's really more comfortable to keep the reference like this.

So the base component is now available but only usable by developers (and only capable of sending message to ActivityPub for now). Things will be really exiting with the next 2 steps, as bidirectional communications will be available, and the gateway will be usable for early adopters. I don't expect those steps to be really long.

test message sent with Libervia AP Gateway

Oh, and to answer another question that I've had, yes you can use the same ActivityPub actor identifier as your XMPP JID. I'll explain next time how everything is accessed.

That's all for today.

debacle 27/09/2021, 08:02

M

matlag 28/09/2021, 00:30

S

sthaan 14/11/2021, 19:48

Libervia progress note 2021-W31

goffi 03/08/2021, 11:03 jabber-xmpp-en SàT Libervia project libre Libervia progress SàT progress

Hello,

last weeks have been exhausting with lot of things at the same time. I've been preparing the release of the 0.8 version, and I wanted to have a couple of thing ready for that, notably a proper way to do translation.

Preparation of 0.8

As you may know, I've implemented a docker integration into Libervia to be able to run relatively easily third party software. This is working, but when testing in the production website I had to put the finishing touches to make it work (notably I've improved HTTP proxy and HTTPS management). I have then created projects and updated a couple of translations files.

As you can now see on https://salut-a-toi.org/, there is a translate menu. Unfortunately I've closed the account creation for the moment, as I have to deal with licensing first. Indeed, nearly all Libervia ecosystem is for now in AGPL v3+, as there are only a few contributors (2 mains one, then only a small patches). The intent was and still is to be sure that the ecosystem stays in an Libre license, but this license may cause trouble in some edge cases, notably if we want to make an iOS frontend (the fruit store is notoriously causing trouble with AGPL licences).

Thus, I'll bring the subject at next general assemble of the "Salut à Toi" association, and see what we should do. One option could be to use FSF's Fiducial Licence Agreement to let the association the possibility to modify the licence as long as it stays a libre one. It would then be possible to add an exception for an iOS frontend. An other would be to avoid totally iOS. Anyway, this need some time and discussions, and if I open translations and get several contributions under AGPL v3+, it may be harder to set this up.

Weblate integrated in the official website

An other time consuming task was to continue with renaming and adapt package names (notably in Pypi). I've used a little trick to redirect legacy packages to the new ones: a new version of each legacy package is a simple setup.py depending on the new package (you can see it e.g. for sat package). I've also put in place a redirection on the Mercurial repositories, using the old repos will redirect to the new ones.

Finally, I've published the 0.8.0 beta 1. You can install it easily with pipx:

  • First install pipx as explained in its documentation
  • Then install the backend with pipx install libervia-backend. You can follow the documentation to see how to configure it and launch it. This will include the CLI and TUI frontends.
  • If you want to test graphical frontends, you'll have to install Libervia Media with hg clone https://repos.goffi.org/libervia-media (assuming that you have Mercurial already installed), then add it into your libervia.conf
  • To install the Desktop frontend, use pipx install libervia-desktop
  • To install the Web frontend, use pipx install libervia-web

Note that the Desktop frontend is still for early adopters, I need to refactor message handling and do some optimisation and stabilisation to make it pleasant to use.

Please send feedbacks either as bug reports/feature requests on the official bug tracker, or on the XMPP chat room at sat@chat.jabberfr.org. I plan to only fix major issues though, as I'm now fully working on 0.9 and I'm focusing mainly on the ActivityPub gateway. However, bug reports/feature requests will be taken into account for 0.9 if not fixed directly in 0.8.

ActivityPub Gateway

After the hard work to move 0.8 close to the finish line has been done, I've started to work on 0.9 and thus the ActivityPub gateway. The first major thing to do was a refactoring of offline storage. Indeed Libervia (or SàT at the time) was started a long time ago with an Async framework (Twisted) long before asyncio even existed. SQLite has been chosen as backend to store data, and a hand made module based on Twisted's adbapi has been created. Despite the rough edges is has been working quite well all this time, and there was even a semi automatic way to update schemas between version. But the whole thing was starting to be difficult to maintain, and the schema update instructions were all kept in the same file.

Fortunately, SQLAlchemy, the most famous SQL databases abstraction with an optional Object Relational Mapper has recently added support for asyncio.

SQLAlchemy is a very powerful and widely used tool, so it has been a quite obvious decision to use it to replace the old system. But to use it, Twisted needs to use an asyncio loop, and Libervia was using GLib loop (or reactor in Twisted terms), due to the dependency to dbus-python.

Dbus-Python is, by its own authors words, not be the best D-Bus binding to use due to unfortunate design decision, so it was the occasion to replace it, and I've moved the backend to TxDBus, a Twisted native D-Bus implementation, which can run with any Twisted reactor. For technical reason, dbus-python is still used for frontends at the moment, but I plan to completely replace it before the end of 0.9 development.

This has required some work, but it was worth it, and after that I could switch to asyncio reactor and implement SQLAlchemy. I've decided to go with the ORM and not the core only as it is opening neat possibilities. I've first made a mapping corresponding to the last version of the database used for Libervia 0.8.

Once SQLAlchemy has been implemented and working, the next obvious decision was to use Alembic, the recommended SQLAlchemy based database migration tools (by the same authors). Thanks to this, migration files are now in separated files, and are really easy to create (Alembic can autogenerate a good part of a script when a migration is needed).

Thanks to all this, I can now easily make changes in database (while in old system I was hesitating due to the work implied). SQLAlchemy also paves the way to support other databases than SQLite. Even if I'm currently sticking with SQLite only, to keep focus, I'll probably add support for PostgreSQL and MariaDB/MySQL at some point.

Once all this work on storage backend has been finalised, the pubsub cache has been implemented.

Pubsub cache is operated transparently for end-user, and stores locally pubsub items (according to internal criteria). This is useful for many reasons: performances of course, but also it allows to do data analyse/indexing, for instance to retrieve all items with some terms (e.g.: to search by categories or hashtags). Pubsub cache is also useful to store data in a component (what is of interest for the ActivityPub gateway), or to store decrypted data (which will be of interest when we'll work on the e2e encryption with pubsub).

I'll pass the implementation details, you'll find the code on the 0.9 bookmark, notably in the pubsub cache plugin, and I've written documentation for developers for some explanations.

New commands has been added to libervia-cli to manage the cache, in particular there is a purge command to delete items according to given criteria, which will save resources and disk space. With it, it's possible to delete only certain types of items (e.g. only blog posts), for all or only some profiles (for instance, only for the AP gateway). You can say a limit (e.g. delete all items which have not been updated for 6 months). Here again, documentation has been written to explain the commands.

While doing all that, I've noticed problem to cache correctly items (because of the flexibility of XMPP Pubsub, it's hard to impossible to say if we can share cache between users), thus I've written a protoXEP (i.e. a proposition for an XMPP Extension Protocol, or XEP) to fix the problem: https://xmpp.org/extensions/inbox/pubsub-caching-hints.html.

I've also submitted a pull request to fix a problem in XEP-0060 (Publish-Subscribe).

While I was a working with standards, I've updated a XEP I've authored a couple of years ago to specify order of items: XEP-0413: Order-By.

Last but not least, while doing the tests for the pubsub cache I've created some helping methods (or fixtures in pytest terms) to help doing unit test.

This concludes the first step of the XMPP-ActivityPub gateway which was, as anticipated, a big one. The following steps should be done more quickly, and work on 0.8 should not be on the way anymore (I plan to publish 0.8 in early September).

That's all for this note, see you next time.

E

edhelas 03/08/2021, 11:08

M

matlag 03/08/2021, 12:50

debacle 03/08/2021, 19:56

Libervia progress note 2021-W18

goffi 06/05/2021, 11:39 jabber-xmpp-en SàT Libervia project libre Libervia progress SàT progress

Hi,

again, lot of things have happened since last progress note, so I'll only talk about major changes here.

"Salut à Toi" is now "Libervia"

The project has been renamed to "Libervia". Even if I personally loved the former name (which was a reference to a French punk band song, an which could be translated to "hi to you", a nice fit for a communication tool), it proved to be hard to pronounce and remember for non French speakers, and the many names of frontends and project components were confusing. The name change has been discussed for long in the association, but the new ActivityPub/Pubsub end-to-end encryption project accelerated things: after a talk with NLnet, we decided to move forward on this so project name would not change in the middle of its development.

After doing a quick poll, we confirmed that "Libervia" (which was formerly the name of the web frontend only) would be the new name.

All parts are now named in straightforward way: "Libervia Backend", "Libervia Web", "Libervia Desktop/Mobile" (currently the same Kivy frontend for both), "Libervia TUI" and "Libervia CLI", with matching executable names (libervia-backend, libervia-web, libervia-desktop, libervia-tui, libervia-cli also aliased as li). The former names are kept internally and as aliases.

The non-profit (French "loi 1901") association behind it stays with the name "Salut à Toi".

This renaming has involved a lot for work, it took weeks to update code, web sites, doc, etc. and according to our statuses, we had to make a general assembly to discuss this decision. It's still not fully finished (notably the official web site URL is still https://salut-a-toi.org, while https://www.libervia.org is currently used for the demo server), and source code repositories have not been modified for the moment, but most of the renaming is done, and you can now reference the whole project as "Libervia"

Official Website Update

Following the changes in Libervia Web themes, the official website one has also been updated and is now based on Libervia's Web Bulma theme. The news now links to my personal blog as it is where you'll have most up-to-date informations about Libervia development (and the former page was broken). Tickets/Bug tracker is now directly accessible from the official site, as it makes more sense to have it there. It's still accessible from goffi.org, and thanks to its decentralised nature, it's usable transparently on both locations.

I've also temporarily disabled account registration on the bug-tracker due a wave of spammy accounts. I will have to put in place a protection for that, but I'm reluctant to use popular non-libre options.

Flatpak and Docker

While working on the renaming and website, I've updated the Flatpaks (they were really outdated), and Docker images. Flatpaks is for now using a specific dev repos, but I hope to see Libervia on Flathub after the release.

I've created Docker images and Docker Compose file to run quickly a local demo of Libervia Web, you can see the instruction on the Official Website.

Ideally, I would like to also create Snaps, Appimages, Nix packages, etc. But I'm lacking time (Flatpak and Docker are already too much time consuming) and prefer to focus on the code rather than on the packaging, help is more than welcome though.

User Friendly URLs

As you may have noticed on the last blog posts, URLs are now more user friendly:

A blog post is referenced using its item ID, and previously a unique ID was used for that, which is relatively long and doesn't give any information about the content, but is necessary to avoid conflict (writing a blog post with an existing ID will overwrite the previous one).

To make it more pleasant, a URL friendly extension was then added, and not used to retrieve the item, so in the example above, www.goffi.org/b/LFMqr7xC2aNf4MDgkbamBY links to the same blog post as www.goffi.org/b/LFMqr7xC2aNf4MDgkbamBY/sat-progress-note-2020-w53. The resulting URL is long and not easy to read, but the item is unique.

The new behaviour directly use URL friendly item IDs, and to avoid conflict, a short random suffix is appended (on the example above, QGqK is the suffix). After some tests, the collision risk for a short suffix like that is not that high (I've tested millions of IDs without collision), and it may anyway happen only if 2 blog posts have the exact same title, so the risk is very low. The resulting URL is more pleasant.

This URL friendly ID is used by default when a blog post is created, but it can be deactivated if user_friendly_id is set to false in blog post metadata, or by specifying manually an item id.

To accompany this change, a new Libervia CLI rename subcommand has been added to li blog and li pubsub, which will change the ID of an item. As there is no standard rename operation in XMPP Pubsub, this is done by copying the item to the new ID, then delete the former one in case of success.

Navigation Helpers in Libervia Web

It was not really easy so far to know where we were in Libervia Web. To help with this, the selected menu is now shown activated, and a breadcrumb has been added.

The breadcrumb is only shown when there are at least 2 elements to show (i.e. not on root pages). It is generated automatically by default, but can be customised with specific label, sub-elements, or even icons, like in the file sharing screenshot below:

Libervia Web 0.8 Breadcrumbs Screenshot

Blog Editor

As it was not possible anymore to write a new blog item with Libervia Web, I've made a blog item editor, which is relatively basic for now, but working. If JavaScript is activated, you'll get a tags editor, preview, and autosaving:

Libervia Web 0.8 Blog Editor Screenshot

File Sharing Quotas

One last missing piece I was needing before release was to put in place quotas on the file sharing component, this is now done.

Indeed, this component doesn't work with a per-file limit like most others do, but with a per-user quota, and you can upload any file size you want at long as you're not over quota.

Release to come

It's more than time to think about the release. I wanted to improve the chat notably in Libervia Web where it's still really basic since we moved out from the former frontend, but finally I've decided to report this to next release, as I plan to refactor messages handling, and for now I need to concentrate on the ActivityPub gateway.

So I'll soon prepare a beta version, and plan to do the release in a couple of weeks. I'll do bugfix on the 0.8 version during this time, but will avoid any important new development.

ActivityPub gateway project

With all the work done above (and other things, I've not mentioned everything), I've been late to start working on the ActivityPub project, but now I can focus on it. The first task is about developing a Pubsub cache as Libervia is currently getting its data for Pubsub related feature directly from the services.

Beside the obvious speed improvement, having a local cache will give the possibility to do data search/manipulation (such as doing Full-Text Search when the Pubsub service doesn't implement it, or doing feature-specific data analysis), handle message received unordered, allow to keep decrypted data when received from e2ee items, etc.

So far, SQLite was used for data storage in Libervia, by using Twisted's adbapi and custom semi-automatic schema update/data migration. It has been working relatively well so far, but it's no pleasant to maintain.

Fortunately, SQLAlchemy has recently added support for AsyncIO, thus it can now be used in Libervia. This is great, as SQLAlchemy is popular and rock solid, so I've decided to go with it. This will open the possibility to use other backends (like PostgreSQL), and refactor Libervia to use SQLAlchemy's ORM. Logically, Alembic will be used for data migration, which should make database modifications easier.

Such a cache will make possible to implements things like items discovery based on categories (or search by "hashtags" as it named in other software).

That's all for this note, see you soon.

E

errormovim 06/05/2021, 11:40

E

edhelas 06/05/2021, 11:59

M

matlag 06/05/2021, 13:40

debacle 06/05/2021, 14:27