Hacker News Reader

Comment by abeiz | Hacker News Reader

The last 12 months have been pretty interesting. My partner and I were making enough off LogicPull to work on it full time. Most of our clients are Legal Agencies in or around Ontario. Due to the sensitive nature of the information collected in LogicPull, it was only possible for these agencies to use LogicPull if, and only if, it was installed on their own servers. This has been an uphill battle from the start. Version control was non existent at these agencies, and any updates and/or bug fixes had to go through a formal code review, which could take up to 3 weeks! We decided to open source our software, and just consult with our existing clients. Its been a great ride!

View on HN

I'm sure this is a byproduct of who your clients are, but the fact that you took the time to catalogue not only dependencies, but to document the applicable license of each[1] is really cool and is the kind of thing that makes it easy to adopt new open source software. Well done!

[1]https://github.com/ChrisZieba/LogicPull#software-requirement...

>My partner and I were making enough off LogicPull to work on it full time

So the struggle was entirely technical? If you were only working on this a year and made enough revenue to work on it full-time, it seems more successful than most startups!

Our main struggle was finding clients to use our hosted version of the software. As we quickly found out, no company wants to have important client information living on a different server. We really didn't like having to create a new branch of our software, then deal with all the licensing involved of it being installed on a remote server we didn't control. In the end, open sourcing was the best option available.

Did you consider going the appliance route? Leasing commodity boxes that can mount network shares via Samba or NFS and reselling them as CPE seems like it would work for this particular software.

That way you would control the software load and be able to manage the quality of the product. And you could use a PPA to manage updates etc.

I'd be really interested to know if you considered this but did not do it and would like to hear the obstacles.

"no company wants to have important client information living on a different server."

Let that be a lesson for the startup people out there.

I work for Bank of America. There's stuff on our servers we don't let other teams in our building have access to.

That's exactly the problem we have at https://commando.io/

No one wants to give us SSH keys to their server. So we have a great product but it's very hard to sell.

"we have a great product but it's very hard to sell" - how can that be true. If it's a great product it would sell.

Obviously nobody in their right mind would hand over their ssh keys to you.

Also the name probably puts some people off. I also don't see any need for what the product does.

"Never type the same commands into multiple terminal windows again!" - I can easily do that already with ssh (or clusterSSH) - I don't want a UI for server commands, I want to script this stuff. When you analyze a little deeper, I think you'll see why the product isn't selling. I hope you are able to make a go of it, but my gut instinct tells me that dashboards and apps like yours are too hard to make into anything substantial.

Well Commando.IO has a hosted version. But I don't believe that rakes Command.IO in much money. If people had more trust, Commando.IO would be covered in cash. It's a great product!

I like how your response to criticism is to ignore it, and say how great your product is in the third person.

Wow. Seem like a great product. Maybe you could license it to cloud hosters? I'd love Linode to have an interface like this.

(off topic) your logo looks very similar to the crunchbang linux logo: http://crunchbang.org/

Founder of Commando.io here. First, I'm not exactly sure who you are, but your not officially associated with Commmando.io. Second, we generate SSH keys, and users then add our generate pub key into their servers authorized_keys. Users can revoke the commando.io pub key at any time from authorized_keys.

Thanks for open sourcing this!

I have to say though, it seems like there are a lot of easier targets for this kind of software outside the legal realm where it wouldn't have to be installed on their own servers.

I don't know about the competitive layout of your industry, but I was wondering if you considered switching into other market segments that wouldn't have things like formal code reviews up front for a product like this.

Why not distribute a vm image with all your stack and then just update the image with new features/fixes?

This was against IT policy of the agencies we were working with.

The open source / consulting strategy may actually prove more lucrative from the sounds of it. Best of luck.

Look. I've worked in enterprise software.

You sound like you did it all wrong.

That you think that 'Version control was non existent' is bizarre is just, well, naive.

Why did your clients need version control? Was your code not extensible?

You do the install. You sell a support contract. You make it extensible in a repeatable, not modifying the core code, kind of way. You have tools to find out if the client has modified parts of the code they shouldn't have.

A bug pops up, you fix it locally. You then log in to their instance. You deploy the fix.

That's how it works.

That's where you went wrong.

It sounded to me like the legal firms wanted to do the formal code review themselves, independently.

How much of the software they run do they do their own code reviews on? That sounds weird (not impossible, just weird).

I imagine it's the only way lawyers could off-set liabilities onto someone else if something else ever did happen where confidential information was breached - and they'd be the best ones to know.

The lawyers I worked with years ago never even thought to ask for code reviews of WordPerfect, and later MS Office or Windows. And they sure as hell knew they had 0 chance of suing MS should anything ever go wrong.

Thanks for taking the time to comment. I wish the process for updating or releasing fixes was as easy as logging into their instance and deploying the fix. We were never allowed near their servers. No ssh, no in-house install, nothing. The instructions for installing LogicPull were documented and then sent off. We were pretty much in the dark the entire time, and there was no way around it.

I haven't looked at the source but I did try the online demo and I too do not understand how upgrades/bug fixes would be so problematic. Or why specific branching for individual customers would be required.

Since LogicPull was conceived to be a SAAS, I can see how certain design decisions may have been overlooked that would make supporting a self hosted solution a lot easier. I also think a lot of people who have never worked in an enterprise environment, do not fully understand how customizable a solution has to be. Politics, quality control standards, etc. all play a role and your solution has to be able to adapt. Unless you are in a position like IBM, Microsoft, etc., where you can advocate certain ways of working, your solution really has to be insanely flexible. And if you take this into consideration, you'll design your solution accordingly.

Reply on news.ycombinator.com