Hacker News new | ask | show | jobs
by hinoglu 5713 days ago
Well i didn't intend to gain attention to advertise my projects, but uhm.. :)

one i'm lately struggling with is visitrs.com, i've posted it a few times before, did not get much response, ones i get was helpful though.It has started as a pluggible chat system which allowed visitors of any website to chat with eachother. It would provide interaction between visitors, would create a community around the website, would help people discuss or suggest products, articles on the website boosting content awareness of the website. for example on ebay, one could have easily ask opinions of other visitors about a product to be bought.. anyway did not gain attention, so i changed it into a stumbleupon+delicious+live chat system. still truggling with it.

one was a real time sql multiplier proxy. problems with production servers back then 2006-2008 was - and most probably still is- backups, cloning on the fly, active-active clustering etc. one would have to stop the working system to backup or clone the datas in sql servers, and active-active clustering were painful or too expensive. I made a working demo based on postgresql + spread real time mesaging toolkit, which was acting like a proxy between sql servers and web applications. when a web application made a query, it was multiplied and sent to all servers in exact same order, which would allow real time cloning + backing up + active-active clustering in once. Though demo was sufficient for simple ops, it needed transaction system implemented, which was really not a big deal. Couldn't sell it. the companies that were paying zillons to big players, or spending thousands of man/day operations did not get what i was solving.

another one that is still on production line is a job listing system, which will get rid of "send cv - wait for response" and "read zillions of cvs, call hundreds of persons just to hire 1" problems. it's a social marketplace for job seekers and hirers simply. i'm not over it yet, but the feedbacks about it is not giving happines yet. addres is test.cbslab.com if anyone interested.

there are many more in the attic, maybe they'll help me when there'll be enough amount of abondoned ones to give up my carreer and become a farmer :)

3 comments

A few experiences with visitrs.

The first 5 seconds of looking at your page filled my head with 'This is just a page of shit that I'm not interested in'. It was not clear that I can participate in creating the list. your join now button is virtually invisible to passing users.

I have to click an 'about' page to find out what the site does. this is way too much effort. this needs to be splashed all over the front page.

Once I understand what it is, it's not as good as my existing bookmarking sites. Struggling to see where the USP is.

My gut feeling is that you might be able to get it to work if you work in the niches - people into baking would rather join a dedicated cake site than something that is as general as this. Buy 10 niche domain names and work in the communities they target. Makes SEO a damn site easier too.

Hope this helps!

I've heard the second sentence a millions times :)

maybe it won't make it, but the best part of it is i got many feedbacks including yours that'll help me in the future.

also i learnt much about erlang, ejabberd, mongodb, js frameworks, xapian etc. not that much of loss i guess :)

I checked our your product. Here are some of the points I find:

1. Your product seems too common place 2. I do not see enough energy in the UI - like stack overflow that looks warm and friendly.. Ask people to upload their photos etc. 3. Whats your niche? Why should I invest my time and not use delicious.

Contrary to what you think, your design is quite neat. Its just you need to select a niche and respond to what your early users are saying!

- Rushabh

ERPNext.com

Though demo was sufficient for simple ops, it needed transaction system implemented, which was really not a big deal. Couldn't sell it.

Transactional integrity across replicas is a big deal--it is, in fact, the hard part. Telling people it's not makes them think you don't know databases and they shouldn't trust your code. Maybe they were wrong (maybe I'm wrong), but it sounds like that's the impression you were giving.

Ops sorry, "replica" is the important part here.

There's no replica in the system. There are clients (clients of messaging system are sql servers) that are not realated to any other sql server in no means, which receive the same commands from a web application in the very same order each time.

Replication is lightweight and done in the sql proxy by replicating the messages, not in the sql servers. Simply it's a 1-1 connection between a web-app and sql-server, but the messages are multiplied by the number of sql-servers which made it a 1-N real time cloning system.