|
|
|
|
|
by account42
81 days ago
|
|
If you are hosting your own git repost you don't really need to provide diffs between any arbitrary revision - just pregenerate diffs between each commit and its parent(s) and tell people to clone the repo if they want anything more fancy. Maybe add a few more cases like diffs between releases if you are feeling nice. And you also don't need to host a version of each file for each commit - those should just be HTTP redirects to a unique URL for that version of the file, e.g. to the commit that last changed it - or just don't provide it at all since most people are only going to be interested in branches anyway and others can clone the repo. The same goes for many other expensive operations that other websites (including blogs and forums) do that cause the website to go down when a bad crawler finds it. It's almost all self-inflicted pain that doesn't even provide meaningful features to real users compared to a better designed website with a finite number of pages that you can even host statically if you want. |
|