Compressing parameters in the URL

Question

The urls on my site can become very long, and its my understanding that urls are transmitted with the http requests. So the idea came to compress the string in the url.

From my searching on the internet, i found suggestions on using short urls and then link that one to the long url. Id prefer to not use this solution because I would have to do a extra database check to convert between long and short url.

That leaves in my head 3 options:

1. Hashing, I don't think this is a option. If you want a safe hashing algorithm, its going to be long.
2. Compressing the url string, basically having the server depress the string when when it gets the url parameters.
3. Changing the url so its not descriptive, this is bad because it would make development harder for me (This is a 1 man project).

Considering the vast amount possible amount of OS/browsers out there, I figured id as if anyone else has tried this or have some clever suggestions.

If it maters the url parameters can reach 100+ chars.

Example:

mysite.com/Reports/Ability.aspx?PlayerID=7737&GuildID=132&AbilityID=1140&EventID=1609&EncounterID=-1&ServerID=17&IsPlayer=True

EDIT:

Let me clarify atm this is NOT breaking the site. Its more about me learning to find a good solution ( Im well aware this is micro optimization, my site is very fast atm ) and making my site even faster ( To challenge myself, and become a better coder ).

There is also a cosmetic issue, I personal think that a URL longer then the address bar looks bad.

Answer 1

You have some conflicting requirements as you want to shorten/compress the url without making it less descriptive. By the very nature of shortening the URL, you will, to a certain extent, make it less descriptive.

As I understand it, your goal is to optimise by sending less over the request. You mention 100+ characters, instead of 1000+ which I assume means they don't get that big? In which case, I'd see this as an unnecessary micro-optimisation.

To add to previous suggestions of using POST, a simple thing would be to just shorten the keys instead of using full names if you don't want to do full url shortening e.g.:

mysite.com/Reports/Ability.aspx?pid=7737&GID=132&AID=1140&EID=1609&EnID=-1&SID=17&IsP=True

These are obviously less descriptive.

But like I said, are you having a real problem with having long URLs?

Answer 2

I'm not sure I understand what's your problem with long URLs? Generally I'd try to avoid them, but if that's necessary then you won't depend on the user remembering it anyway, so why go through all the compressing trouble? Even with a URL of 1000 chars (~2KB) the page request won't be slow.

I would, however, consider using POST instead of GET if possible, to prettify the URL, but that's of course depends on your implementation / environment.

Answer 3

It is recommended a few times here to use POST instead of GET. I would strongly recommend AGAINST picking your HTTP action by what the URL looks like. There is more to this choice than how it is displayed in the browser.

A quick overview:

http://www.w3.org/2001/tag/doc/whenToUseGet.html#checklist

Answer 4

A few options to add to the other answers:

• Using a subclassed LinkButton for your navigation. This holds the extra data (PlayerId for example) inside its viewstate as properties. This won't be much help though if you're giving URLs to people via emails.
• Use the MVC routing engine to produce slightly improved URLs - no keys for the querystring. e.g. mysite.com/Reports/Ability/7737/132/1140/1609/-1/17/True
• Create your own URL shortener like tinyurl.com. Store the url in the database along with each of the querystring values to lookup.
• Simply setup some friendly URLs for the most popular reports, for example mysite.com/Reports/JanuaryReport. You can also do this using the MVC routing engine.

The MVC routing engine is stand alone and can work without your site being an MVC site.

Answer 5

With my scheme, one could encode the params section of a URL as a base64 string which is ~50% shorter than a direct base64 representation. So for your case you get:

• ~50% shorter params section
• a base 64 string which hides a lot of the detail

see http://blog.alivate.com.au/packed-url/

Answer 6

Most browsers can handle up 2048 characters in URL; if you don't feel like to use a long parameter list, you can always to pass parameters through POST requests.

Answer 7

There are theoretical problems with extended URLs. The exact limit varies across browser (roughly 2k in sort versions of IE) and server (4-8k in Apache, varying on version and configuration), and isn't officially specified in any RFC that I am aware of.

I would agree with synhershko, and replace the URL with form POST parameters instead if you are concerned that your URLs are growing too long.

Answer 8

I've encountered similar situations in the past, although my reasons for optimisation were for SEO. To me it depends on what you're doing with the page URL variables, are they being appended on all/most pages? If they are then to me there is almost always a much better way, although if you're far down the development path it's probably too late now.

I like being able to 'read' a URL, especially when I drop into an unknown site 2 or more layers deep in the navigation and there site is designed poorly, it's often the easiest and fastest way for an advanced user to find where they are on the site.

If you're interested in it from an SEO point of view, its normally best to have a hierarchy which only contains: / - _ Search engines will try and read URL's, see this video by Matt Cutts (can't remember how far into the video he mentions it but it's a good watch anyway...)

Answer 9

Any form of compression of the URL (hashing, compressing, non-descriptive) is going to:

1. make the urls harder to read, remember and type in correctly
2. have a performance impact as you will have to decrypt/decompress/convert the url before you can work with it.

Also, hashing is usually considered to be non-reversible - given a hashed value you shouldn't be able to work out what generated it, but you could use it to look up a value in a database, which gets you back to your first issue of short-long lookups.

You could easily just remove the redundant "ID" at the end of each parameter, and possibly strip out vowels or similar to "shorten" the url without losing too much from the semantics of the request.

But to be honest, the length of your URL is one of the least things to worry about in terms of performance - look at the size of any cookies you're sending back and forth between the browser and the server, and the page size you're sending back.