The Design of Software (CLOSED)

A public forum for discussing the design of software, from the user interface to the code architecture. Now closed.

The "Design of Software" discussion group has been merged with the main Joel on Software discussion group.

The archives will remain online indefinitely.

Yet another searching tool (What build...)

Before the explosion in desktop search tools like Google Desktop, MSN Desktop, Copernic, X1, etc... I started a program to locate information stored in diferent files.

Then the explosion come, and everybody is doing it (yuck!). I have now the understanding of how is done tecnically (in fact, I start - a port in Delphi of Lucene, with the posibility to run - later - in Win32/NET/Mono/Linux plataforms) however this is only the tech side.. I want to know how about the bussines/desing side.

This product is comercial. Is NOT a *desktop* search tool, I hope build something usefull for a whole company (ie: Search across the entere network).

My company is located in Latin America/Colombia so I don't face a high competition in this arena but at least, I want compete in the whole Latin America market and need support of english/spanish and maybe portugess languages/interfaces.

I want go version 1 *quicly* (in less 1 year from now). I hope get the engine (MUTIS, the open source part) in some months (I decide do the hards parts first, the rest is easy for me).

Apart for common things in desktop searchs tools like crawl documents (office, multimedia, texts), provide a web-google like GUI, search of mails (outlook) and that stuff I plan this things:

- "All in one": I can embeb the Web server/database/any inside the engine, so no tecnicall people is necesary to operate/mantain/patch a web server like Apache/IIS (good for small pymes) and provide the modules to plug into IIS/Apache for the rest. No aditional licenses or software about anything related with this product.

- "Secure by default": I can provide a SSL comunication with basic encryption and compresion, again, ideal for small pymes (that its, the search engine is exposed in a web service and the communication is protected with your own SSL/Digital firm). Posibility to use normal/enterprise options later

- "Fail over/Fault tolerance free": The product not need nothing special to operate in array, the server can auto-discover backup server/load balancing servers inside the LAN.

- Integration with Windows security (Domain controlers). I think this is hard... (that is respect the security and mimic the logic of Windows security...)

- A tool to admin the search repository and all this stuff, change/install new ways to ranking or document filters, etc...

- Connection to any database accesed with ADO/ODBC/Native drivers (like Sql Server, Acces, Firebird, oracle, etc...)

- "Schema virtualization" : For example is defined the customer schema, with the classic fields. Then part of the data come from a ERP database and the rest from Outlook, then I merge the data and provide a unique view of this... in a single record (not duplication). All the findable data in this project is modeled around a scheme (for example exist the file schema, the image schema, etc...)

- Multiple GUIS: I hope provide a web-based GUI and alternative guis for Wap/Office.

- SDK: The ability to change the engine (for that is mutis open source) and build integrators, filters, etc...

- Distributed spider: The spiders can be deployed to any computer and communicate with the central/array of search servers

- The hard: I need cover the small pyme (ie: from 1-15 computers in a company) with almost zero technical experience (the norm) but be able to go at least to 50-200 sized companys...

However, all this stuff is so tech ;)

So, taking in account that desktop search tools and some enterprise tools are now common and potentialy a lot of people know about it, in what more think?

And is easy to see that do all of this stuff can be complex, in what FOCUS for version 1?

What SUCKS in actual solutions?

Thanks for your time...
Mario Alejandro M. Send private email
Monday, May 16, 2005
Try a viral approach.

Start with two users as the base case.  User1 wants to share documents with User2.  User1 right clicks on a file to  start your app - your app then indexes the file and adds it to a local repository which User2 can now search or browse.  User1 and User2 are now sharing and searching documents.

Enter User3.  He wants some documents from U1 and U2 so he installs your app. He can now search/browse User1 and User2's documents as well as share locally selected documents as well.

So and so forth until you have created a distributed/searchable high value file network.

This approach would help sidestep corporate security hawks and the problems of a centrally managed "be-all end-all" search solution rolled out by management that no one wants to use.

Once your get 10 people in the office using your software you can try to sell them db integration and that sort of thing.
lumberjack Send private email
Monday, May 16, 2005
Ok, I understand...

I'm playing from some time with the idea of permit single users to download the tool free but display a small text (like AdWords) where can say something like "Get this for your office too!) or something like...
Mario Alejandro M. Send private email
Tuesday, May 17, 2005

This topic is archived. No further replies will be accepted.

Other recent topics Other recent topics
Powered by FogBugz