3 ms·
I think the author got a little too caught up in flickr's cleverness of making an ID based on different information. For instance, the first letter is the post
by chacha102 16y ago
I think the author got a little too caught up in flickr's cleverness of making an ID based on different information. For instance, the first letter is the post type, then a group of letters is date, and then there is the post number for that day.
The fact is, not many people are going to go devise their own URL shortening algorithm and implement it. While it is a good idea, I just don't think many people will put in that kind of effort.
What might work is a more general algorithm. An algorithm that can reliably shrink and unshrink any URL.
I'm not sure, but this might not be possible considering that many people have invented compression algorithms, and the general rule is that if you compress data, you are going to loose data. (I'd like to hear someone's opinion on that considering I haven't dove deep into compression)
Edit: By 'possible' I mean 'meaningfully possible'. Taking a URL around 50-80 characters and making it around 20-30.
- tel 16y agoI am not an expert, but for lossless compression I believe the idea is that if there's a mismatch between the size of an encoded symbol and it's information content (novelty) -- then you've got some room for lossless compression. Is that the case with URLs? Probably not directly. Certain atoms (.com, www) might be encodable to save space, but it'll not necessarily be a big gain.
- nitrogen 16y agoIt might be enough to shorten only the most common sites and platforms' URL forms.