Hi Sandy,
> I am reaching out to get a sense of how extensively PDF certificates
> are used by other Open edX instances
As disccussed in Slack, Stanford uses these extensively.
> and how much of a burden it would be for those instances to move to
> web certificates.
Too much. We have no plans to do so.
Besides the effort involved, we prefer the UX and product workflow of PDF certs.
> edX.org no longer supports PDF certificates
Stanford has maintained our fork of the codebase since late 2014 [1].
Before we forked, we were the active maintainers of the core repository;
we used to have commit access and our developers have authored 78/136 of
the commits to-date [6].
Since forking, we have continued to improve the codebase, including the
introduction of a YAML-based templating system which significantly
increases speed and ease of certificate design and development [2].
> as we move away from using RabbitMQ, we are also moving away from
> supporting PDF certs because they are integrated directly with
> RabbitMQ
This isn't quite right; certs integrate via XQueue, not RabbitMQ [3].
> they are not managed by Celery like other asynchronous tasks.
There's some truth here. The certificate "agent" [3] is a
supervisor-managed service that polls an XQueue endpoint and
pops/processes jobs from there. We've previously expressed interest in
decoupling certs/xqueue for varying reasons, namely that we have a few
deployments that otherwise don't need XQueue and it would be nice to
remove that dependency from the infrastructure.
I did a proof-of-concept of this waaay back at the 2015 OpenEdX
Conference Hackathon, but never had the time/motivation to make it
production-ready [4][5]. We'd be interested in finding a path forward
in revisiting this. That said, unless edX has plans to deprecate XQueue
(do you?), this seems to be a less pressing concern.
> We are looking at the possibility of deprecating PDF certificates as
> part of the next Open edX release and possibly even deleting the code
> entirely.