jtdaugherty / mathblog Goto Github PK

Is there any chance that a new version of mathblog is going to be available soon that supports the most recent version of Pandoc?

Having the Mathjax header embedded in the program obviously makes it difficult/impossible to be customized for the user's needs.

Right now it's impossible to specify an alternative config file. That would be really, really nice for a production-testing distinction, for example.

Post generation can take a while if a post has a lot of generated images (e.g., TikZ). And generating a whole blog (e.g. with mb -l) can take long enough to be annoying. So add support for running a (configurable) number of post generation processes concurrently.

With ghc 7.10.2 and a fresh cabal sandbox, I'm seeing some indication a dependency has changed its API:

src/MB/Util.hs:108:36:
    Couldn't match expected type ‘time-1.5.0.1:Data.Time.Format.Locale.TimeLocale’
                with actual type ‘System.Locale.TimeLocale’
    NB: ‘time-1.5.0.1:Data.Time.Format.Locale.TimeLocale’
          is defined in ‘Data.Time.Format.Locale’ in package ‘time-1.5.0.1’
        ‘System.Locale.TimeLocale’
          is defined in ‘System.Locale’ in package ‘old-locale-1.0.0.7’
    In the first argument of ‘parseTime’, namely ‘defaultTimeLocale’
    In the expression: parseTime defaultTimeLocale "%s"

src/MB/Util.hs:115:32:
    Couldn't match expected type ‘Pandoc.Pandoc’
                with actual type ‘Either
                                    Text.Pandoc.Error.PandocError Pandoc.Pandoc’
    In the expression: doc

Integrate a lightweight web server into mb so the user doesn't have to run a separate web server alongside mb -l.

Right now it's way too easy to check out a copy of your blog, update a post, regenerate the content, and find that the 'posted' timestamp is now entirely wrong, since you lost your original filesystem state. The original post timestamp should be stored in the post-index so it can be preserved and/or modified by hand if desired.

At present the RSS feed includes only the page title because the page content uses relative URLs to reference generated images (such as LaTeX math), which won't work in an RSS feed. However, generating content with absolute URLs is unfriendly for blog testing where one wants to run a local instance without having to worry about the particular production URL that the blog will eventually use.

It's possible to have mb use a different config for 'production' or 'development', but it isn't possible to force absolute URLs on all page resources.

So: we need a way to force absolute URLs on everything (i.e., generated images, next/prev post links, the post index, stylesheets, etc., etc.).

I had to run md --force to get updated pages after changing my gnuplot preamble.

The fact that the Blog type knows anything at all about GNUplot preambles is an artifact of the old implementation in which GNUplot was more tightly integrated. This needs to be changed so that only the GNUplot Processor has to worry/know about the preambles directory.

This does suggest there needs to be some relationship between the blog skeleton and the configured document processors, though. One idea is to add a field to Processor like generateSkeleton :: Maybe (FilePath -> IO ()) which gives Processors an opportunity to take part in the skeleton creation process.

I would like to use both Tikz and Gnuplot in my blog. It appears that I can configure only one of the two backends globally. Is there a way to specify the back-end to be used on a per-post basis?

And, to take things one step further, can I mix Tkiz and Gnuplot graphs in the same post somehow?