Tumblelog by Soup.io
Newer posts are loading.
You are at the newest post.
Click here to check if anything new just came in.
If I'm free to throw some ideas of what could have been done better  (for discussion and extension, maybe it would be a base for cleaned up requirements version in the wiki):

* not using Ruby on the backend ;-)
* storage and duplicate files will be a thing; maybe S3 or Backblaze would be a good option, though I don't know what exactly current soup does.
* duplicate files detection, this happens on soup a lot. Not sure if we should MD5 uploaded content or maybe do some visual comparison magic (and pick the better quality on top of that?).
* GIFs should always be converted to videos. Save bandwidth, save quality, no scaling. Go gfycat-like!
* webp whenever supported!
* Oh, speaking of which, not sure if the downscaling done by soup is needed, sometimes the files with _500 suffix are bigger (byte-wise) than the originals
* when things fail, retry/remember. This is done currently for  unfinished posts, but not for reposts.
* the way tumblr does reactions (as adding to the original post) sometimes seems appealing...

* an API for the mobile app. Oh yes, and a mobile app.
Reposted byRekrut-Kkelu

Don't be the product, buy the product!

Schweinderl