Same goes to the job postings. This is an example, posted in a developer group somewhere.

Mandatory – 4+ years experience in PHP, MYSQL, JQuery, Ajax and CSS
Advantage – Experience in Ubuntu, EC2, Drupal, Perl and JAVA
Advantage – BSc/BA in computer sciences

MYSQL? No, MySQL. JQuery? No jQuery.  JAVA should be Java.

In my mind it is less excusable in a job posting then it is in a CV.  Company has more resources to get its marketing materials right. And make no mistake, job positing is as much a marketing material as a CES poster.  The quote above screams ineptitude.

I’ll not mention “degree in computer sciences”. It is computer science in which one gets a degree.  But this is not as easy to verify by searching Google for 20 second .  Getting  jQuery spelling right is.

Very good meetup again. I enjoyed it. What made it ever so entertaining that people on stage assumed that developers in the audience are all about to start up their own company or product, or already did so. Even more entertaining, that from a quick poll done by me in the lobby, the assumption is probably correct.

While liking the meetup, and acknowledging the points made and welcoming the response of entrepreneurs in the audience, I want to try and look at this from the pure developer enjoyment point of view.

Assume I were a salaried software engineer coming from traditional Java or .NET or C++ background. Strike that, lets stay with C++. This will provide more of an equal footing. Or if I were a university computer science major,  thinking what advance classes to take. In a series of posts, I’d like to review each platform, so that to give re-training developers and university computer science students the developer, not business, overview.

I am going to quickly look at iPhone, Android, Windows Phone 7, WebOS, J2ME and Blackberry development. I leave Symbian aside because it is being phased out, and beyond that I am yet to see a person who considers it a match to other platforms in terms of fun to work with.

If you program in Python or Java, and do not need SQL database,  use Google App Engine. It is free until you have heavy traffic. Prototypes rarely do. It allows for both Java and Python code to run. GAE, as Google App Engine is popularly known,  has had it share of uptime and other issues in the past. That made some developers leave GAE. Warranted or not, it could only be the issue for a production system. For prototype it is as great is it comes.

If you develop in ASP.NET, use DiscountASP.NET. They are not free, but 5 USD/month. It was 10 when I used them when I was into ASP.NET. I used them  for almost all my clients. They rock. And a lot of developers seems to like it enough to make it community choice year after year.

I am not sure what to use for PHP. There must be a lot of choices out there. Please comment and recommend.

SOAP is XML and XML is verbose, if compared with JSON. SOAP schema id verbose on its own right. It is also very complex. JSON-RPC and REST/JSON are simple, succinct,  easy to implement even without any framework. Messages are easy to parse and process, that is cheaper to host and easier to scale. That is why I choose REST/JSON internally.

SOAP however is great for IDE integration. SOAP is so well formalized, in fact, well defined to be formalized, that tools support SOAP integration with ease. This is expecially true for .NET and Java.

Derive  a class in Visual Studio from certain base, add some methods, mark them with an attribute, you got a SOAP web service.

Include a remote Web Service as if it was a local library, provide the URL to it, and presto, you have object browsing and code intellisense, as if it was a local library and local objects.

That is why to lure developers in Java (the most popular programming language at the moment) and C# and VB (to other popular languages) I feel the need to expose my web services as SOAP as well as REST/JSON

And let me confess, when I  develop my server side as ASP.NET or Servlets, SOAP would probably be my first choice. Nowadays I tend to choose Python/Django, so my first choise is REST/JSON. But let me assure you I don’t forget to quickly follow up with SOAP alternative.

If you’d agree that eventually having both flavors (or all flavors if you really need 3rd parties supporting your product) is necessary, the question is, how does this knowledge changes your design of the API.

I shall dedicate a post to it in the future.

My current spot is on PHP vs Python. Imagine this. I run a small start-up company. It builds a web based product. The product is good in it own right, but allows other people to write plugins to it. Some plugins are web based as well. Some are mobile apps. Some are desktop apps. I want to encourage 3rd party developers to build plugins, because it cements my ecosystem.  My developers are good. They are fast to learn and can do either PHP or Python or both. They can do Ruby, Java, .NET or Perl. They can do anything. What is the setup that is ideal?

First there is nothing particularly wrong with either. You can be successful doing web dev with either.
Goolge is heavy on Python. Facebook is a PHP shop, for the most part. So either language is fine. But fine is a dirty word in the quest for the best.

Python advocacy is here http://www.infernodevelopment.com/python-becoming-most-popular-programming-language.

PHP has a lot going for it. It is easy to get started with. It is easy write web UI in. PHP is popular. That means readily available workforce. TEOBE ranks PHP having twice Python popularity, on par with C#. There is Eclipse configuration build specifically for PHP, but not one built for Python. Since my hypothetical company is web, PHP seems to be the perfect match.

Python is said to be more “elegant”. It is difficult to quantify. Disregard this argument. Wait. Python 3.0 is not compatible with Python 2.5. While this is pain in the ass, because Python 2.5 had accumulated a lot of useful libraries, the willingness to start client says something about what language designers care about. Python is thought out language. We know who designed it, we know what make them to design it like this. Decision to break with the past is not something that is done lightly. It could have done because the past was a failure. That is why we have Windows Phone 7 incompatible with Windows Mobile 6.X. Python 2.* was and is popular. Decision to start fresh has to do with the relentless quest for elegance and well, perfection. So while weither or not Python is elegant, there is an attempt to have it that way.

The killer for me, however, is that Python is versatile. PHP runs on the web server. Python runs on web back end, on desktop and on Nokia s60 phones. It comes preinstalled on Macs. It runs even in the browser (Iron Python in Silverlight). It scripts 3d game engine Panda3D.

This versatility means that python  is not just for the core web applications for my company. It is for reference code for plugins. It can serve my system administrator instead of csh scripts. It can be one language for everything.

C# is like this. Java is like this. PHP is NOT like this. May be I shall choose the best tools for each platform and HTML5 will end up being better then IronPython Silverlight for my web plugin. But if I find myself in a multi-platform situation, especially when I need to create an ecosystem, I find Python a convincing choice.

Web server code is difficult to debug. The fact I can debug most of the code in a desktop application is a huge benefit for me.

HTML5 brings a number of things that makes it a better choice for many. Here is my list of those things.

• CSS3 effects. RIA is all the more about eye candy as it is about functionality. CSS makes it possible. Round corners and transitions and gradients and transparency are key.
• Offline. No need having Gears. HTML5 is almost as capable of storing data client side as Flash or Silverlight.  I wish there was standard for “structured data” aka SQL based storage, but blob is good enough for most needs. Storing JSON in blob is good enough.
• Javascript libraries. jQuery and other libraries have matured. A number of jQueyr plugins is mind blowing.
• <audio> and <video> tags. Right, there are incompatible between browsers in terms of formats. But there are good player libraries already that work around this issue, failing bakc to Flash for the older browsers.
• iPad and iPhone. iOS devices famously dis Flash, but Silverlight does not fair any better. iPad sells in millions but its importance on technological choices people make is well beyond what sales numbers suggest

On the other hand Silverlight got a new life in Windows Phone 7. I shall review this situation in the upcoming post

My own very humble opinion, open not open, who cares? As a developer the only thing I care is the return on my investment of time and money developing for a platform. That is why I never cared if Windows is not as good as Mac.

This is not to say I have my own aesthetic preferences. I think  Silverlight running C# is more elegantly designed then Flex.  I think Flash should have used pure javascript not ActionScript.  All that is so remotely second to me wanting to make money off my software that I only mention it here to make a point. Which is, I’ll see where the market is, and code in whatever language it takes.

Apple makes great devices. I am not going to argue Android vs iPhone.  Android might be great too. But Apple makes hit after hit after hit. Apple knows what users need. So users  reward Apple’s efforts by buying those devices.  Developers choose Apple because users like Apple, not because Apple likes developers.  Apple does milk 3rd party addons, tunes and apps and accessories alike.  That is fine with me. It is their domain and therefore it is their rules. I can choose not to join, but if I joined I shall not be complaining.

In that light, let me say, I welcome Apple dissing Flash. I know HTML and I know javascript. I don’t know Flex almost at all. If Flex dies, Flex developers will likely move to HTML5, making its acceptance in browsers faster, tools better, books on the subject better written, samples of code more readily available. And I shall not have to learn Flex.

I guess I was lazy. Oh may be I realized that nobody reads my stuff, so I might as well save my time.

Now I decided to relaunch. The new reason for blogging is that my memory is not what it used to be. This is going to be the log I write for myself. I shall keep it public because some people might find it useful.  Making it helpful, however, is not a goal. It is a byproduct at best.

Therefore I don’t put in any effort to make it readable.  So some posts will be cryptic. Some would have abbreviations only I would understand.  Sorry for the inconvenience.

Security. It is mostly a psychological problem then real one. Your data is probably safer on other people servers then on your laptop you’ll leave in the airport or taxi. But the psychological problem is a problem none the less. And I have no clue how to help you fight it, other then saying : I keep my data online and I am not afraid.

Usability of web apps – speed. Most web apps are slow. The problem is a compound one. Browsers are slow to interpret Javascript. Javascript, if you don’t know, is the probramming language your browser must process for that feel of resonsiveness that makes best web apps feel like desktop. Network is slow. Some things can not be done inside your browser, so a web app talks to its server over the internet. This conversation is relatively slow.  Older web apps used a page model, where every new thing required you moving from one web page to another. Those web apps feel slower. 

Solution to the above is likewize compound. Browsers are getting better. Google Chrome 2.0 and FireFox 3.5 are speed demons when Javascript processing is concerned.  More and more people get broadband access.  Newer apps are written using a trick called AJAX, and you stay on the same page while Javascript brings new stuff from the server. It is not faster necessarily but it makes you feel it is faster. And because of this constant communication, your web app brings things you don’t need this instant but might need after a minute, and by then they will be ready for you. That what makes Google maps feel so fast.  New Javascript libraries appear that make web app programming so much easier. Then we got Silverlight 2.0. Silverlight 2.0 is Microsoft technology to allow in-browser applications to be written in C# using .NET framework. It is revolutionary because Silverlight runs code that is compiled and therefore it is faster then Javascript.  Silverlight is not available for Linux and it requires installing a browser plugin. I have high hopes for Silverlight to enable new responsive applications. However I don’t see it happening yet.

Usability – need to log in. Your desktop does not require you to log into every application. Web apps do. It can’t be avoided, your information out there is not secure without it.  But without some help, the necessity to log in every time, to remember all those passwords because a huge obstacle. I privatly think it is the main obstacle, though most commenters place security first. Anyway there are tools to help you. First that comes to mind is password managers built into most browsers. Besides many web apps allow you remember the session (not log out). That is good if you use your personal computer most of the time and nobody else uses it. Which is not the case for most potential web apps users. It if were, why would we want use web apps in the first place?  Solution I like it password filling plugins, of which Roboform is my personal favorite. It is not free, and there is no, to the best of my knowledge, a free alternative. But I find having it important enough to pay for one copy per every computer I actively use. You can find my post about Roboform elsewhere in the blog.

Data portability. It is very hard to copy data from one applciation to another. If I want to attach my Google Spreadsheet to a project on Basecamp, I have to go to Spreadsheets, save as xls file, then go to Basecamp, then upload. It is incovenient for one file, it is madness for 10, it is impossibility for 40. And I have no good news for you in this department. I don’t know of a good solution.  There are some standardisation efforts for inter-application exchange, like OpenSAM, but I am yet to see two good application I am ready to use that can exchange data using that standadard.

There are other, smaller problems, but those, in my humble opinion top the list.

I use Roboform on two of my machines.
These are my main working machines.
I can’t image not having it on a main machine of mine.

Roboform is a browser plugin that fills up login data.
Your passwords are encrypted, and you only have to enter master password once.
Firefox, for example, can store passwords, but Roboform shares them across my IE and Firefox,
and I can copy them to another computer and use there.

I can even copy them onto my HTC smartphone which has pocket Roboform installed.

I have grown to be dependent on Roboform very much.
Google Chrome, which was very good in version 1.0 was no go for me, because it did not have plug-ins, hence no Roboform.
Google Chrome 2 has “extensions”. I am waiting for Roboform extension.
Then I’ll be able to make Chrome my primary browsing platform.
I can live without every other Firefox add-on. Roboform I can’t live without.
Fortunately Roboform vendor works on Google Chrome’s extension mentioning alpha in 2009

I have two gripes about Roboform. One is that to synchronize my passwords on two computers I have to copy “passkard” files.
I would expect some kind of web service, a central repository with which I’d synchronize.

Two, I don’t like it costs money. It worth the money all right. But why such thing is not free?
There is a lot of great free add-ons. Why not Roboform knock off?
Primary free alternative is KeePass (http://keepass.info).
I did not try it recently, but every 6-9 months I make a quick search, and so far, no, there is no alternative to Roboform.

The key factor is one-clickness. It is almost as smooth as letting browser remember your password.
You pay the small price of extra click for password portability.

Buy Roboform for every machine you use, hope for Server sync product from them and home that KeyPass people will get serious about copycatting Roboform excellent user experience.