Hprc banner tamu.png

Talk:SW

From TAMU HPRC
Jump to: navigation, search

20170510/j-perdue

Preliminary examples (strictly HTML) available at:

  1. http://hprc.tamu.edu/software/ada/
  2. http://hprc.tamu.edu/software/curie/
  3. (terra currently down)

Things still to do:

  1. in retrospect, having a seperate index/listing doesn't make sense. Leaves the index page rather devoid of content. Instead of a separate listing.html with names/descriptions, I think it would look better if it was all just below the string of letter links on the index page
  2. the cluster description at the top could be improved upon. At the very least, add a link to the user's guide to the cluster.
  3. need to convert to PHP and find a simple way to add in our header/banner2/footer includes just to give it more color.
  4. possibly update tables to use HPRC CAS for formatting (I got little clue on this but can try to copy other examples)
  5. add a timestamp to each page that indicates when it was last updated
  6. setup nightly cron scripts to update and push to web server

20170508/j-perdue

(note I have no idea of the best etiquette/formatting on this but will run with what I have and see what happens. BTW, I realize some of our users or others around the world might see this, but.... maybe they have some ideas...)

I haven't found any "simple" solution for providing a software listing on the web short of "module spider", which isn't a solution.

I've played with some of the possibilities.

I really think we can do better with less work.

Currently pondering something like this:

  1. create a set of web pages for each cluster (each in their own directory, e.g. hprc.tamu.edu:/var/www/html/software/ada) that include:
    1. in index.php - a brief description of the environment (HTML from a static file on the cluster e.g /sw/local/etc/ada.html)
    2. in index.php - a link to a brief list of available modules (just name and description) with <A HREF=...#NAME> links (generated dynamically nightly in a seperate file)
    3. in index.php - an index {[a-z],[0-9] to a list of subpages (updated nightly)
    4. in a bunch of .php files... for each [a-z],[0-9] above, create an alphabetical subpage that includes a table with the needed <NAME=> bits (if that can be done in one table). For each module/application, list/include:
      1. the name, wrapped by its homepage URL
      2. the description (again, so they know they picked the right one)
      3. the versions available
      4. any notes/links on how to use it (e.g. links to our wiki page like SW:Abaqus). (all updated nightly)

The terse mode of Lmod will be vital in this. e.g. "ml -t"

.php is probably preferred over .html to slap some stock HPRC headers/footers in there if we want.

The trick will be trying to extract the "homepage" and "description" from the EB modules instead of the HPRC modules when possible since the latter don't necessarily format them the same as EB [which should be the primary source where possible].