[i336_]

Some additional details about the project may be useful to others - you've described a fairly wide range of scope at a rather abstract level here.

I'm quite confident a number of highly knowledgeable people have looked at your question but have been unable to answer for lack of context. Narrowing things down a bit would likely get you more focused answers.

With that said, the best tech stack is the one you already have in your pocket. You could argue that this is why having a good all-around knowledge of all of the major stacks is a good idea.

[AgharaShyam]

Thanks for the suggestion i336_

So what i am planning to build is SAAS service that provides all the feature's listed above and more, customised for an (unspecified for competitive purpose) industry. The web client is used for entry of data related to sales, stock, payroll. I am also making apps for various operator's inside the plant to upload data using their android phone's and tablets.

Then i want to process and display all this data (related to Manufacturing, Sales, Marketing etc) on web, iOS and Android.

So essentially there will be apps for employee's working in the plant(for data entry) as well as for the managers and owners(to monitor everything and make decisions)

I want the web dashboard to look somewhat similar to - https://demo.baremetrics.com

And since i have no experience in building mobile apps i want to know what languages/frameworks would be best suitable to develop good/native looking apps for iOS and Android using web technologies.

Thanks.

[Matthias247]

That sounds like something that would often be built on top of a SAP stack. Or if not directly built on top of at it would probably need to have integration points for it.

[joaocbatista]

have you checked outsystems? it seems to fit your needs.

[i336_]

After carefully considering what you've said, my first question is what development experience you do have. On the one hand the "finished product" overview you've provided could simply be for conciseness, or it could be lack of experience. (My point is that I have no idea.)

On the surface you're describing something that's been done many many times (kinda obvious). On the one hand this means you could find someone to work with to help you on this fairly easily, on the other hand you're looking at 5 figures easily... which is likely why you're trying to figure this out yourself, now I think about it :)

I'm hesitant to suggest looking at existing off-the-shelf solutions; I get the impression that finding systems that haven't rusted together in the areas you need flexibility is touch-and-go and hard to figure out in advance. It may go incredibly well, or you may end up maintaining a massive headache that's incredibly hard to fit-out for your needs. One possible idea that may work to some extent is finding open-source projects that implement some or all of the functionality you want; they may be useful for parts/ideas or real-world answers to "how did they solve this bit," or even better you may find they actually work for you (yay free stuff!).

While you may come up with extra defining/deciding context/requirements, the main decision point I'm thinking of is whether you consider it a must (or a huge-value add) for things to be partially functional if there's no consistent network connection. This applies to sales/management (on a train, or at exactly the right vortex point between a bunch of buildings, and cellular just dropped out?) as well as plant operators (hypothetical contrived example: a non-essential machine just went bezerk and is spouting so much RF the Wi-Fi just died). That said, generally supporting transient network loss is... hard; you need a database design that, at every point in read/write, can tolerate "another client may have committed an update but that data might not be in the query we just did yet," which the application design must be built with awareness of (it's so fundamental it can't be abstracted).

If "the plant Wi-Fi just died" disrupting operations is okay (for example, you have somone on-site who can rapidly diagnose a wide range of issues, you have spare APs and related kit available, etc), then you'll save yourself a lot of time by just developing this as a webapp. Others may argue for native apps, but if you do a webapp, you can target desktop (Win/macOS/Linux(?)) + mobile (Android/iOS) just by using media queries. Suddenly several thousand lines of Swift and Java disappear :)

With enterprise setups like this, you also get to call the shots on what you contractually agree to support Android/iOS-wise; you get to specify the minimum versions, and you can also check the user agent and throw warnings or blocking error messages for versions you don't like. This isn't a public Web site, so you don't have to work for everybody. And so you get to guarantee that certain Web APIs are available :)

In all fairness there is the AppCache and ServiceWorkers+IndexedDB can cache web apps for offline use, but I wouldn't rely on them because the storage areas they use can be cleared.

Now I think about it, you could build minimal Android/iOS shells that load your HTML/CSS/JS into a WebView with some simple caching tacked on to guarantee code availability in offline situations. That could work quite nicely, and the native codebase would be tiny. With this Wi-Fi dying isn't a problem but now you need to go back to mentally modelling eventually-consistent database schemas.

On the backend side, what sort of traffic volume do you intend to handle? Do you intend to have the backend be on-site? (You could do a managed on-site setup where you have (monitored) SSH by contract but the system runs on client servers.) This would distribute processing load across the sites. If you want to centralize everything then you need to think about how many clients you want - then double or triple that figure (and then maybe multiply it by itself) - and you should have a good idea of the sort of volume you need to be able to process. There's a reason all the "how we process 1.21 jigabytes of messages every second" posts on here are so popular! You may need to look into high-throughput, easily-shardable datastores.

Regarding those graphs... well... note that you're looking at the revenue the guy who built those graphs is getting! :D I can see there's a fair bit of intelligence in the graph displays, and fine-tuning a similar system with a quarter of the functionality is something that would take a bit. Certainly possible, but it would take a bit.

Unfortunately I have very little advice to offer in the area of actual Web development; I haven't explored the scene much yet. I know that there are a dizzying number of developer offerings available. Your best bet is to forge ahead trying them and see which one you like.

I'm not a developer myself (in that I don't have a software engineering position - nor the mindset for one, just yet); I just spend all my time on HN :)

I wonder how much more specific you could be (for the benefit of everyone) without giving anything away.

By the way, thanks for mentioning the Baremetrics demo - that's really cool that they're open like that.

[AgharaShyam]

Thanks for a very detailed answer. I appreciate it.

You are totally right in that i don't have any real web/application development. In fact i just started university but i have been programming for a few years (mostly problems on sites like code chef in C, C++ no products) but recently got this idea when i visited a plant that my father owns. There are about 600 such plants in the same area and they use fairly old software or not at all. So i talked to him about my idea and he thinks it will be very useful and also commercially successful since there is virtually no competition. That's why i decided to learn web development. But now i am realising it's tougher than it looks. I asked this question in order to get an idea what all the things i need to learn in order to even attempt to build something like this. Now that i look at my question i think i should have asked "How to be a full stack web developer?". So if anyone is still reading i think this would be the question that you should answer or maybe i should have just googled it and started working

[saluki]

Nice that you have a connection to get your foot in the door at your Dad's plant. Customer #1.

The path I would recommend to learn full stack development.

Go through the books Head First HTML and CSS (get an inexpensive hosting account you can FTP files to and setup some domains, point to your hosting), next go through Head First PHP and mySQL. Next is Headfirst Javascript then Headfirst jQuery.

After those you should sign up for teamtreehouse.com.

You could play around with developing this in PHP without a framework, setting up a login in system you would learn a lot. Sometimes frameworks are tough to understand what all the magic is happening behind the scenes if you haven't creating a simple app with login, etc in vanilla PHP on your own. For your customer version though you will want to use a framework, more secure, better organized code, etc.

I would recommend developing this with Laravel(PHP) it has a great community so you'll be able to find great learning resources to get started. (Rails is good too, but I think Laravel will be easier to get started/deploy your app)

Also I recommend developing on a mac. Windows always seems to have obscure issues that you will waste hours/days on.

laracasts.com will get you started with Laravel.

There are great deployment tools forge and envoyer. Hosting on Digital Ocean is perfect for most apps, AWS and linode too)

Use stripe for billing.

There is a SaaS add-on to the framework call Spark. It's ok but, I find I'd rather roll my own stripe/SaaS integration.

Also Spark does use Vue.js though so since you're just learning I would stick with developing the app using jquery for ajax interactions and think about moving up to vue in the future.

I would create the MVP as a web application for all the interactions. Then once you have that established you can look at adding true android and iOS apps. Those might be better to hire out the development of. Or a future learning project.

You should be able to set everything up as a mobile web application, use the bootstrap framework for the look and feel of your app. There are some ready made admin templates that can get you something similar to baremetrics.

Having your Dad's plant as an initial user for your app will be great to refine what they need get their feedback.

Good luck. Follow up and let us know how it's going.

[i336_]

Ah, interesting! Well, you've done some homework already and defined the problem area quite well, so you have that going for you.

C and C++ aren't the easiest languages to learn given they require manual memory management, so you have it fairly easy learning something else, unless you actually want to use C++ for the backend (I would hesitate a lot to recommend C). The main advantage of using something like PHP/Python/Ruby/Node vs C++ is the nonexistent compile times, along with the fact that they're all much higher level and let you focus on your app's requirements. That said, no language is perfect; it's about picking the one you can live with the easiest. Another potential consideration is to choose a language that a lot of programmers in your area are familiar with, so that you have a competitive market to take advantage of if you need to bring in extra help.

I'd say that it would be significantly easier to not spin this a general-purpose SaaS stack, but instead build this as a bespoke system specific to these plants' requirements. It will make developing the system significantly easier, especially given that this would be your first major project. Practically every enterprise/environment is full of custom applications; these are not unique at all.

(That said, from an educational standpoint, thinking about how you might generalise what you build would be a good idea, and serve you in building similar systems later on.)

Assuming you haven't already done this, what you definitely need is information about exactly what everybody needs to be able to do. You have an absolutely excellent opening with the fact that your Dad owns one of these plants: you should be able to easily get access to wander around your Dad's factory and ask questions of everyone when they have moments of free time. I would strongly recommend having long conversations with the workers themselves and completely hearing out how they would solve the issues they have on the floor. Potentially be prepared for the application to suddenly become very different than what you planned :) - you may end up electronically enabling/adding brand-new functionality to these plants' capacity/capabilities alongside the analytics and number crunching features you initially started out with.

It might be a good idea to do several rounds of questions, as you can discuss things in greater depth as you internalize what you learn and come up with unknowns as you work out how to turn problems into helpful code. You may or may not find asking questions helpful at the point you're at right now. Maybe an idea to discuss with your Dad.

As for an immediate long-term goal to figure out how to reach, I'd suggest taking the architectural requirements you have thus far and going ahead and figuring out how to put together a system that provides those things. This might be an existing system you buy (although I would strongly recommend grabbing some freelance consultant help for an independent go/no-go if you think about this route) or an open-source software stack that you can start with, or a stack you build yourself from scratch.

You may need to go through several iterations of designing, spending long days and nights on building, and then realizing you have to throw away half the system and start again. This is not a bad thing; in fact right now I'm building a crawler for a specific website that I'll use once and throw away. It's gotten to a couple thousand lines of code right now (it's quite simple, but PHP is quite a verbose language for async network stuff, and I'm unfortunately yet to learn a better language like Go or Python), and I'm basically putting it together for this once-off use-case. I've mentally saved a LOT of valuable information about Web crawler design that will go into my next major project, a more general-purpose crawler (note the things about generalization I mentioned earlier). I couldn't get this info from reading tutorials or even academic papers or theory.

Oh, another thing - databases are weird and annoying, but I would highly recommend you learn them in depth. There are lots of choices; MySQL/PostgreSQL are the de-facto relational options (and for MySQL there's phpMyAdmin, which is awesome for mental modeling and simple setup); Cassandra scales out to a staggering level; SQLite3 is great for learning on and prototyping with; MongoDB comes with utterly insecure defaults and doesn't scale well but does kinda work (not dissing it, just being honest). There are many others. Query plan optimization is something that should be absolutely logical but is full of obscure edge cases, and writing the same query with different syntax sometimes gives inexplicable speed boosts. Lots of confusing fun - my point is figure it out to the point where you're comfortable with databases and can easily reason through them.

You probably discovered http://stackoverflow.com/ long ago; if you don't already have an account with a few hundred reputation I would recommend going on there and answering a few questions so your account has enough reputation that you can do things like use chat and such. Some of the other Stack Exchange sites like http://codereview.stackexchange.com/ may also be of help.

Good luck, and I kinda want to hear how this goes :)