Update process

[seandavi]

It looks like the "isa" package is not available and has been archived on CRAN. That breaks our ability to build the workshop materials at this point. Any suggestions on a reasonable process to fix issues going forward?

[mtmorgan]

Interesting. It seems like there are two strategies. Create an archive of packages at the time of the conference (packrat, etc; docker; the AMI), and always use those. This is great for reproducibility but misguides the user about what they can expect in their own installation. Or let bit-rot set in, and then it would be good to have an automatic way to flag chapters as no-longer-functional.

Experience trying to maintain code from Bioc books was not encouraging -- it became increasingly difficult to find alternative implementations, and the code became increasingly fragmented into evaluated and not evaluated code chunks.

[lawremi]

I agree with Martin.

We can't be expected to maintain every workshop presented at the conference. Probably best to separate the workshops into tiers, with some kept up and others persisting only in the preserved environment. We'll need to maintain the recurring workshops anyway.

[seandavi]

I think we might want to consider something like #54 to enhance modularity and make it easier to have individual authors troubleshoot. This may move us away from vanilla gitbook, but the benefits of separating content from process (of compiling materials) may make it worth something.

I guess it is obvious, but without some continuous integration, including chapter-level checking, of some kind, the bit-rot will go unnoticed and errors are hard to catch.

Perhaps @LiNk-NY has some thoughts, also.