Hacker News Reader

Comment by dkhenry | Hacker News Reader

While Microsoft might be a good solution for many people (Use what get the job done), this specific article has absolutely no real reasons why you _should_ use Microsoft. The truth is if One 8-gig windows server is what he needed for his start up he could get by with one 8-gig Linux server and save a couple of bucks a month. While that might not be a big deal now, if he ever wants to get big, those little costs start to add up.

Also while its great Microsoft is subsidising his starup via the Biz Stark program, we have seen many times what happens when you rely on "free" to run your buisness. Even tually he will need to start paying for those software licenses he is getting for free and by then he might not have a choice in what platform he is using.

View on HN

>"this specific article has absolutely no real reasons why you _should_ use Microsoft"

He does give some reasons, which might not apply to everyone is another thing: (from the article)

    Because I love C#.
    Because "MS is expensive" is a myth. 
    Because of Visual Studio

Of all the three I am very agreed with the third one. I wish it would be different but I haven't used a better alternative to VS so far (XCode, Eclipse, Netbeans, etc. just didn't work for me for several reasons )

Wow I hate VS so much, Eclipse is so much better! I guess it's just personal preference.

I like C# a little more than Java, but I cant stand assemblies. Jars are much easier to handle.

edit:

Troll? Not.

Like a very simple thing: quick watch in debug blocks the edit window? Really WTF?! I have to move it the second screen to see the actual code. I can name a lot of these small annoyances.

I can't help but feel like you're trolling. Eclipse is a slow, bloated, feature-laking pile of crap. I honestly can't believe anyone would prefer it to, well, just about anything else. I've used it and I get by on it when necessary, but in terms of preference its at the bottom of my list.

> "Eclipse is a slow, bloated, feature-laking pile of crap."

You should try Kepler. Some amazing performance improvements out of the box.

Rubbish. Use the latest version with an SSD with 8GB ram and it flies - SSD and ram are cheap enough to not be a barrier to entry, either. It's got more features than any other IDE I can think of, as well as a pretty huge library of plugins. Sounds like you're the one that is trolling. If Eclipse really was that bad, nobody would use it, and it wouldn't be probably the most widely used IDE in the world.

>If Eclipse really was that bad, nobody would use it, and it wouldn't be probably the most widely used IDE in the world

I don't think this means much. It has (poor) support for just about every language under the sun and its free so of course its going to be widely used. It has many features for Java, but features for other languages quickly fall off a cliff. But its the slowness that is a showstopper for me unless I absolutely must go with eclipse. Yes, eventually computer technology will advance to the point where its bloated codebase is no longer a liability, and perhaps we're approaching that. I use the Juno release occasionally for python programming and everything I've mentioned still applies.

Personally I'm not a big fan of plugins in my IDEs: they always feel slow and clunky with a "bolted on" feel. This has been true of every IDE + plugin environment I've tried so far.

>Troll? Not.

If you had said IntelliJ I might have believed it.

Excuse me sir or ma'am, do you have a moment to talk about the Good News that is IntelliJ? :)

And the list seems to be missing the most important point.

    The .Net Framework is extremely powerful and extensive.

The number of problems the framework has prevented me from needing to solve is staggering. We get work done here, fast.

Eg. Recently I had to convert times from any random time zone into another random timezone accounting for daylight savings etc. I cannot imagine having to solve that problem in just about any other language suggested in this comment section.

Anything with access to jodatime (or nodatime, for that matter) would find that a pretty trivial problem.

In my experience you get more built-in with .net, but richer third-party ecosystems elsewhere.

I agree, he could easily be using a JVM-based stack to get the same or better efficiency out of his 1-server configuration.

Also there's some clearly naivety when bragging about his "low cost" $200/month setup. His SQL Server licenses alone will probably triple or quadruple his costs when his Bizspark status runs out.

Well, AFAIK he can easily switch to MySQL for instance, under the same ORM.

MySQL is possibly the single hardest database to port to/from, because it supports nothing.

In that case porting FROM it should be simple. TO it would be hard if you're using all those fancy features of your current DB. That's why I eschew all the sirens from Oracle DBA's of "oh... but Oracle can do that too!" Keep database activities in the database and application activities in the app.

In my experience, MySQL lets you get into a state where questions like "hang, on - what's the difference between a DATE of '0000-00-00' and NULL?" are commonplace. This makes porting from it... not simple.

Although I have no experience with MySQL, this kind of problems have to be resolved by using LINQ to SQL since every db type is automatically mapped to its .NET counterpart. The DateTime .NET type then has only one representation of null.

Or what you're saying is that if you store a date of zero VALUE to MySQL it could be magically replaced by NULL internally? That would be really crazy.

I seem to recall it is actually the other way around, and storing a null results in a nonsense date.

Added bonus: some versions will segfault if you pass an un-formattable date value.

The problem is that MySQL deviates from the standards (all of them) in surprising and uncomfortable ways. porting from MySQL is fun, when, for example, your target database actually ENFORCES the constraints that MySQL accepted but silently ignored. MySQL doesn't have that feature, but it won't bother to tell you.

Postgres, Oracle, MS SQL all support a (very large) common subset of features, so if you avoid stored procedures and db-specific features, you can port among them easily. This is not true for MySQL.

>"I agree, he could easily be using a JVM-based stack"

Maybe but believe it or not, there are people who HATE Java (that would includes me)

I'm not a huge fan of Java either, but sometimes it's important to get past that and either try and work with a language that's extremely effective for some deployments, or consider that the JVM supports quite a bit more platforms than just Java.

There's always Scala, clojure, groovy, etc.

Oh you're not alone. I dislike Java as well, but there's no denying the power of the JVM and the ecosystem around it. If you really dig into it and look at it as a polyglot platform with choices like Groovy, Clojure, JRuby, and Scala, it becomes more palatable.

Regardless this post is not about .NET tooling, more about server efficiency and cost. Two areas where I think the JVM is a little further ahead.

To be fair, the post is titled "My start-up is Microsoft-based..." not "Here's why your start-up should be..." I feel like he just wanted to make a fair case that Microsoft isn't impossible to deploy as a viable solution for a start-up. In start-ups versus say...a corporation , there is a write-off of Microsoft pretty much out of the gate. I agree with most companies, there are certainly a stack of reason why not to use it. But, its nice to see someone putting some personal reasons out there why Microsoft tech isn't useless to a new, fledgling company.

Reply on news.ycombinator.com