Smarty

[boots]

Here's a question:

Legal issues aside, if you do use a web-service in your app (for any purpose) what is the best/most thoughtful way to incorporate it? Should you have the client issue the web-query (originates from their IP) or should you originate the query from your server? In both cases, should you actively cache query results? Where is the caching more effective?

I think there are reasonable answers to these questions that mitigate a lot of concerns. I assume that there is a legal difference between redistributing data which you acquired via a query from your server vs having the same query run from a client machine.

Technology blurs some issues: When I buy a translation dictionary and do manual lookups, this looks the same as the technique above that does it automatically. The difference is that the technique above essentially permits me to "share" my "copy" of the dictionary removing the normal assumption that all users will have their own copy. I don't think information is the type of thing that should (in my mind, COULD) be owned but I fear I am in the minority on that position. People want to be paid endlessly, often for things they didn't even themselves accomplish.

[toma]

[boots]

Actually Toma, if I try to sell a translation dictionary that turns out to be derived from another commercial translation dictionary--that's illegal.

You're right: I only posted this in the translation thread because it seemed that service issues is where that topic was going anyhow Anyhow, here it is in a new topic!

[toma]

Yup, but it could be argued that a translation file is part of a dictionary, which is just silly.

As for web services:

I have many people that query my site. I have provided external web services for years. I also use web services other sites have setup just for my use.

Where to cache data depends entirely on what data the service provides.

First example: I provide a service which returns a wddx packet containing concert information from my database. If the site querying this data wants their own copy of the data for long-term use, they should cache it locally (on their end). If the person using the service will only occationally look at the data, they should requery my service each time they wish to see the data. This allows regular updates or fixes to be made on my side and, because they only query it occationally, it's not a burdon on my site. Example: http://db.etree.org/wddx/show.php?artist=gratefu&date=2/9/73

Second example: A service which provides a list of all records in a table on my site. This should always be cached by the requesting client and, perhaps, refreshed once a day or less.
http://db.etree.org/wddx/artistlist.php

Third example: dynamic data one one site to be mirrored on another. I use a service on another web site. The data from that site changes daily but the service puts a high load on the server. Therefore, I only query the other site once daily at 4:20 AM (a very non-busy time).


I've found web services to be a fantastic way to share data between sites and, in my case, I don't concider the data 'mine' (although it could (poorly) be argued it is), therefore, I love sharing my data as much as possible.


My normal web services implementation: client provides parameters in get request and I return a wddx packet with their results.

[andreas]

[boots]

I'm already offtopic here--

You do a haul at 4:20 ? !! That's rich Anyhow, its cool to know that you are the one running etree -- that's a good thing, even though I haven't traded in years (and was never exceptionally active) many of my friends still do. Does that mean you are a canuck, too?

I'll post something ontopic a bit later...I'm going to a show tonight

[toma]

andreas: altavista doesn't provide commercial translations. Read the page linked to in your copied text. Also, my plugin doesn't provide translations for commercial purposes. HOWEVER, if you were to implement a page (probably easily done by using some of my code) which allows users to type an arbitrary string and it used babelfish to translate that string and return it to the user, that would be disallowed.

boots: Yup. backups have been running at 4:20 since etree and, later, db.etree's inception. The server team is commited to following the rules we laid out long ago.

The only canuck that was ever on the server team was Scott aka ghost, but he's been off the server team for a very long time.

[andreas]

I would like to know how good/bad the different translation-engines work, so here 2 results:

[boots]

Now wait a minute--I started this topic because it was off-topic (more general) than the ongoing translation topic.... now we have the reverse problem...

@andreas: Can you please repost your translation related item in the translation topic? Thanks.