Odds and ends

Post #21,783 by ChrisR 12/17/01 10:49:46 PM Reply	Odds and ends I had an idea for dynamic web forms that can handle different languages. Sounds like Todd's field of expertize. The curve for language translation is definitely not linear - the first 75% is relatively painless, but the remaining considerations get more and more difficult. Of course, squeezing out those last few percent is where the money is at. I have an idea for Dynamic Data Caching for a POS system. I interviewed for a job earlier this year for a company that does retail POS systems for gas and convenience stores - a bunch of ex-EDS engineers were involved. Generally speaking, there solution involved using some OO techniques where they developed their own scripting language that embedded the transaction processes within the data. The data was acting as a state machine carrying it's own instructions to complete the process. Hence, if the machine got detached from the central database, the data still carried enuf transaction info to complete the process. There were certain instances, though, where the transaction could not run without verification from the central repository. Specifically, some company credit cards have constraints about what can be purchased on the cards (only a certain amount of diesel fuel per day - and definitely no beer). So even though it was intelligent enuf to complete some transactions in isolation, it had to be aware of scenarios where it had to refuse to process transactions based on the system state. POS is kind of a nice mixture of embedded systems combined with standard business processing software (inventory, pricing, etc). The embedded processes include interfaces to a multitude of gas pumps, registers, credit card readers. The business processing includes interfaces to a bunch of different mainframes and servers with software ranging from Cobol, RPG, VB, Java, etc. The key to development, though, is like any other software - find a customer who is both willing to be a test bed and simultaneously willing to put money into the project development. Kind of hard to get the product out the door if you don't have experience with a paying customer - not to mention money comes in handy in getting over the rough spots.
Post #21,804 by tuberculosis 12/18/01 1:58:55 AM Reply	Re: Odds and ends RE: Multi-lingual web content Sounds like Todd's field of expertize Yep. Here's the thing. You don't want to store whole html web pages as translatable elements. The reason is that people who do language translations don't know jack about html - so they invariably delete a closing tag here or there. Instead, you need to classify tags into two types: structural (table, paragraph, list) and inline: (bold, link, italics...). Structural tags are what your templates contain. Its your page layout (although you need a way to have locale specific layouts for the people who read right to left and top to bottom). Then you put all the text in as large of blocks as possible into your database, inline markup and all. The reason you keep the inline markup is that it tends to move around. Example: English: The red [link\|http://www.apple.com\|apple]. Spanish: La [link\|http://www.apple.es\|manzana] roja. Notice that the link locations also change. So links want to get handled specially too - one more level of lookup needs to be kept to translate links (or not - depends). Writing an intelligent html editor that can handle all this freaky stuff is hard. One more thing to keep in mind is that translations can usually be done on a sentence by sentence basis. Some phrases recurr often. For instance, suppose you are translating the website for the Gap. The Gap has trademarked the phrase "stone washed jeans". One additional tool is translation memory - a phrase database that stores translated stuff within the context of the current content. A good translator's editor will tie into the translation memory database and actively try to assist the translator with known past translations. This helps provide decent consistency if you have a team of translators working on a big site. The translation memory database is shared and the editors are tied in "hot". Its more like groupware. There's a bunch more issues wrt font rendering and character encoding conversion that needs to happen in the web server but suffice it to say that almost any other idea is a better one than trying to get into multi-lingual web content management. FWIW, my former company, after hiring a total moron as VP Eng, and allowing him to eliminate my job after I already designed everything and built most of it, has finally fucked me by doing a 640 to one reverse stock split in their last round of funding. Basically diluting me from maybe getting my boat paid for to making it unlikely that I'll be able to buy a cheeseburger after any IPO. I'd prefer they'd folded the company so I could just write off my loss.
Post #21,853 by nking 12/18/01 11:47:53 AM Reply	Err not the whole HTML just the parts of it that the database will fill in, like the title part, the header information before the input box, etc. The ASP page or CGI program will parse out the HTML and fill in the blanks with the data from the database table. "Oy! Seats taken mate!" - "Brilliant!" on BBC America

Welcome to IWETHEY!