nicholasdavidson
Generalize pybits?
I'm a developer with @Connexions, an Open Education Repository, housing CC licensed textbook-like materials. Basically, authors worldwide create content on our site, using a defined XML (cnxml), which we then serve as HTML pages, PDFs, and EPUBs. We're in the process of rewriting our system, one part of which is the 'exports', the scripts that run when content has been published (or republished) to generate the PDFs and EPUBs (as well as some zip-file bundles of "source"), which are then pushed back into the repository.
If we map content repo -> distro source, buildd -> export build scripts, it seems pybits matches what we've designed for this. How tied to buildd do you think pybits is? Would this use-case be interesting to you? Basically, you're using the same tech-stack I was planning to use (PostgreSQL, RabbitMQ, Python), so the question is how do I treat my fork of your code, as a "something upstream will never be interested in" so I'll tear it apart, or should I attempt to generalize and make pluggable bits?
Ross
It depends what you mean by buildd . pybit was designed so you could stack up multiple build-clients on a single machine.
The way pybit was written it distributes packages of a given type to a particular build-client (buildd). This doesn't necessarily have to actually compile anything, it just requires a PackageHandler impliment the checkout, build and cleanup stages. These would be easily adapted to running an epub build or similar.
We would be quite happy to accept anything along this line or your general comments on architecture if you think there is a better way.
Currently pybit is quite architecture focused e.g. it's queues by default look like Distro-Arch-Suite-Package. For a specific version of ubuntu this would be: ubuntu-i386-precisepangolin-deb.
When you don't care about lots of that there may be a bunch of our stuff which is just overhead for you.
Yup, that sounds like how I was planning to tackle it. Thanks
Ross
On Fri, Nov 16, 2012 at 3:24 AM, Nicholas Davidson <[email protected]
wrote:
The easiest way of handling this would be to treat each book type as a different package type, and write a class which extends PackageHandler to build your various kinds of book and to. E.g. check out xml source from VCS, build epub. We would be quite happy to accept anything along this line or your general comments on architecture if you think there is a better way.
— Reply to this email directly or view it on GitHubhttps://github.com/nicholasdavidson/pybit/issues/35#issuecomment-10441089.
Ross Reedstrom, Ph.D. [email protected] Research Scientist phone: 713-348-6166 The Connexions Project http://cnx.org fax: 713-348-3665 Rice University MS-375, Houston, TX 77005 GPG Key fingerprint = F023 82C8 9B0E 2CC6 0D8E F888 D3AE 810E 88F0 BEDE