These are chat archives for biojs/biojs

23rd
Jun 2018
Megh Thakkar
@Megh-Thakkar
Jun 23 2018 08:30
@rowlandm I'll send a diagram as soon as I get one but for the time being, biojs-sniper is working the following way:
  1. The components have the dependencies listed in the package.json file with the key "sniper".
  1. The urls mentioned are either cdn urls or relative paths. If they are CDN, they are kept as it is. If they are relative, they are completed with respect to the repository so that it can be served using rawgit.
The problem comes when the corresponding scripts themselves have further dependencies. As mentioned by @sarthak-sehgal , one of the visualizations of cytoscape requires a JSON data file, mentioned in its js file.
This is when sniper-utils comes into picture
Before rendering it, it will substitute all these links with their absolute paths.
I can modify and provide @sarthak-sehgal with the js files already present in the repo/cdn.
But if I want to modify the links in the js files and then provide it, I will need to serve them from our end, which means saving more data in the database. As the JS code for the same exists, we can directly pick it up. So, we will try this approach for now and see how it works out.
rowlandm
@rowlandm
Jun 23 2018 15:28
You can serve the files directly from the file system. ie save the git hub repos down and serve them directly using nginx.
Nice summary by the way too @Megh-Thakkar !
Megh Thakkar
@Megh-Thakkar
Jun 23 2018 17:35
Yes @rowlandm I thought of that. The only issue is I will have to keep track of the changes in package.json and update the served files accordingly which means downloading them when the cron job runs! I found an integration between CDN and rawgit which already serve files according to every branch!!! I will use them and Sarthak will be updating the URLs in the final scripts, wherever required.