Provisions for the Journey

  • Jun 19, 2016

In my last post, I ended up recommending that Domino developers and administrators take some time, if at all possible, to dive into new tools. In some cases, that may be in order to stay indefinitely, but, even otherwise, learning a new environment would have tremendous benefit to your existing Domino work.

Since I'm not a full-time administrator myself, my advice is admittedly a bit thin. It mostly involves my oft-repeated suggestion to try out a reverse-proxy setup for HTTP. That could parlay into use with mail as well, but even just becoming more familiar with a multi-tiered setup and the essentially-standard OpenSSL tools for certificate manipulation would pay dividends.

For developers, there are a great many potential paths to walk, and I think it's extremely useful to at least take one step down each. The use of that reminds me of the "Programming Languages" course I took in college. It involved rapid-fire coverage of C, C++, Scheme, and Prolog - four distinct languages with pretty distinct programming styles (the latter two moreso than the former two), meant to introduce students to the different notions of programming. I don't use Lisp or (lord knows) Prolog in my day-to-day programming life, but having learned a little bit about functional and logic programming at the time has paid dividends over the years.

There are a couple broad topics I can think of that are worth investigating, including ones that we as a community have already been working on.

Source Control

For the most part, the importance of source control has basically sunk in as a "we should do this" sort of thing in the Domino community. However, the inconvenience associated with maintaining an on-disk project means we're not all the best at doing this religiously. And, indeed, we're often hampered by both old habits and Domino's licensing terms that prevent us from doing this properly: there's a strong strain of "multiple people working on the same NSF" in Domino's history, and that doesn't mix well with source control (or XPages, for that matter).

So I recommend not just knowing about source control, but also trying (as much as licensing allows) to use this as an opportunity to switch to a more "traditional" model for app development. I like to think of this type of development as equivalent to working on a desktop app: it would be madness to have multiple people working in the same Xcode/VS/etc. workspace on the same binary build, and then particularly trying to have an end user also use that resultant binary. Instead, it's better to have each developer work separately on distinct non-replica NSFs, ideally on local dev servers, and then deploy the changes as an NTF.

Java EE WARs

Java EE in general is a very worthwhile target for interest, especially because it encompasses so many other possibilities. But I'd like to call attention in particular to the WAR file model, which is roughly equivalent to an NSF for app deployment.

When I first learned web programming, with PHP, it followed the "just toss a bunch of files on disk and visit their paths in a web browser" model, which basically worked, but taught some very bad habits. Domino improves on this a bit, with everything in its place in an NSF, but still suffers from a bit of the same malady: the "filesystem" of design elements is what is exposed to the browser directly, URL-wise. In a more fully-app-based system, though, there's a lot more flexibility about how the incoming URLs map to the underlying code and resources, and Java EE's encapsulated nature plays on this well.

With a WAR file, you have the "app container" advantage of an NSF but also gain the "everything after the app name is mine" benefit of a controlled URL router, and that shift helps conceptualize apps as something of a "mount point" onto an app server. So you may have a blog app that you load up as "/blog" on your server, and then that app can control the traffic beneath it, but is similarly a bit walled off from everything above it. It's the sort of "feel" that settle in with time, and then it feels weird to lack that flexibility.

This model also fits particularly well into app platforms like BlueMix or Heroku.

Testing

This one's difficult for Domino developers! No matter the tooling used for an NSF, being able to dive into automated testing is difficult. You can coax Maven into running some tests, though that will lack an XPages environment. You can test on the XPages side with JUnit4Xpages, though that won't be part of an automated build without more hoops. There's no real equivalent to the type of fully-fleshed-out testing tools available for other platforms now.

So that's lead to unit/integration/whatever testing not even being on the horizon for most Domino developers. That's particularly a shame because it's already something of a chore to get into the testing habit even when the tools do everything they can to help you. It's essentially like brushing your teeth when you're a kid: you do it because you're told to, but it's not something you'd ever do of your own accord.

It's worth it, though! It will probably require working on a fully non-Domino project, but going through the steps of building a fleshed-out system with automated tests along the way is a valuable experience. The goal is to get to the point where it feels weird to not have tests, much in the same way as lacking source control - an untested change should feel like it doesn't "count". It's a big world, too: beyond just small unit tests, there's tooling to run a web app in a test environment and run user interactions, or do the same on mobile-device simulators.

Build Servers

This dovetails with testing, in that one of the primary benefits of having a build server like Jenkins or Bamboo is that they can act as a neutral, clean environment to build the whole app and run the tests, informing you of any errors or failures. This is another thing that's made much more difficult by Domino, though it's possible.

Beyond testing, having a build server reaps benefits with project organization and delivery pipelines. Jenkins's recent updates have focused on that sort of thing, where the tool will help you manage pulling down the latest code from SCM, building it, and then deploying the results to dev or live servers, as desired. Build servers won't necessarily change the way you develop, but they will prove to be a faithful companion, with more uses the more you get comfortable with them.

Unix

Get comfortable with the command line. Other than Microsoft's tooling, which has a very schizophrenic relationship with the CLI, basically all modern development environments swim in the Unix world. Not everything has to be done via CLI - certainly, it's much more comfortable to use SourceTree than to do all git commands manually - but it's hard to find a tutorial that doesn't have at least a few tossed-off terminal commands. Mac- and Linux-based developers are at a distinct advantage with this, but Windows has Cygwin and the ported Linux userland in the next Windows 10 release.

Along this line, it's tremendously useful to have a bit of Linux server admin knowledge. I recommend either spinning up a Linux VM locally (Ubuntu has served me well) or using Linode. Linode in particular is so cheap and convenient that it's very much worth tossing them a couple bucks to run a server for a little while with full confidence that you can melt it into the ground without consequence. The delightful iOS-focused podcast Under the Radar had an episode on this topic a while back that's worth a listen.

You should probably also learn a little about vi.

MVC

Back in the realm of pure programming knowledge, it's extremely valuable to learn about MVC and related concepts. XPages sort of brought this to the Domino world, but it suffered from lack of wholehearted adoption and its legacy baggage made it very difficult to really go full MVC even if you wanted to.

The good news is that essentially every other active framework encourages or enforces this approach. When diving into one, it can be a bit jarring at first - there's a lot of "ugh, can't I just put the code here?" going on - but it falls into place very quickly.

So where to start?

Picking a place to start is a daunting prospect, given the ever-shifting sands of the modern development world.

One of my favorite tutorials for a web framework is Michael Hartl's Rails tutorial, which is kindly available for free online (with options to purchase other forms). It not only provides an up-to-date introduction to Rails development, but also covers a lot of the other ground mentioned above: the command line, Heroku, Git, and test-driven development.

In another direction, I recommend taking a look at Meteor, which is something of an all-in approach to JavaScript-based development. It does an interesting job blurring the lines between client and server and makes a very compelling case for "JavaScript everywhere".

Beyond that, really just learning any development toolchain will help. There's not too much need to sweat picking "the right one" at each step - any knowledge gained in one will help in others, either directly or by giving you some additional perspective. The most important thing, especially now, is learning something.

Change Bitterness and Accidents of History

  • Jun 17, 2016

It's pretty easy to see that change is in the air for Domino types. It's been taking a number of forms for a while now - the long delay since the release of 9.0.1 and associated aging of the tools and infrastructure have led to a series of forced adaptations for developers and administrators. Developers, for example, have had to keep light on their feet to adapt to new browsers and devices that the framework doesn't automatically support, as well as a shift toward manually including jQuery and other tools that have a bit more wind at their back than Dojo. Administrators, for their part, have had a series of heart attacks related to SSL and other security matters, usually involving a a lot of noise followed by (unfortunately, I feel) a good-enough patch from IBM.

That sort of thing isn't likely to get any smoother. In large part, that's entirely distinct from anything IBM does: the genie's out of the bottle when it comes to fast-moving platforms, and the best we can hope for is a sort of still-moving linga franca that can be sort-of-stable on the majority of targets. But then part of that is Domino: for all its virtues, it hasn't adapted for the modern world, no matter how much some of us would have liked it to. And that's had some negative side effects on us as a customer base, side effects that manifest as a gut-reaction rejection of the modern ways of doing things.

Take the SSL thing, one of my soap boxes: though the immediate problem could be summarized as "IBM should update their SSL stack", the larger issue it exposed was that our beloved monolith is mortally vulnerable to a single component falling behind. And, in fact, it made clear that we're spoiled by the approach: many of the reactions were basically that we shouldn't have to worry about things like reverse proxies or the general notion of distinct systems for web front-ends and the app server. And that initial rejection of the hassle has implications, limiting the average Domino installation's capacity to scale for load balancing or failover in a smooth way, things that come almost "for free" with a reverse-proxied setup common among almost all other app servers.

Developers have it worse: the "power user turned developer" history of Notes and the particular peculiarities of the platform* have left us desperately behind the baseline for modern development. The tooling has coddled us into preferring inline scripts and procedural programming to structured code organization, into viewing a thoroughly-staid language like Java as something outrageously complex, of only begrudgingly adopting SCM due to the platform-induced hassle, and of almost entirely ignoring automated testing. And similarly to the SSL discussion, the historic "one tool for every job" nature of Domino leads to natural pushback when faced with other platforms.

And I get why! And I feel it too. It's a real PITA to now always have to be on the lookout for some new point release of iOS or Android to break drop-down boxes or something, as opposed to years of deploying Notes apps that looked and worked identially across every version forever (pretty much). It's also a drag to run into situations where the problem is "easy" on Domino but more cumbersome elsewhere, like platforms that farm out their FT indexes to distinct servers, or don't include document/record-based security. That makes it very easy to become blinded to the tradeoffs, though. It may be nice that Domino is a one-stop-shop for so much, but it's a shop that requires Designer, that makes it very difficult to use third-party-dependency systems like Maven (even within Maven projects), that lags in DB features found elsewhere, that is only awkwardly accessible from other app-dev frameworks, that has an API that's a bit older than Windows Me, and that essentially never showed up for the modern development conversation.

Domino has always had a lot to recommend it, and XPages has carried us very far. And hey, this is enterprise software - even if there's never a major new version, there'll be paying work forever. It just may not be the kind of work you want to do, and it is almost definitely not truly healthy for the companies paying for it. So what I recommend is that you have a plan. The good news is that there are a great many next steps that build smoothly on existing Domino knowledge and, potentially, infrastructure. Certainly, I'm thoroughly biased in the direction of Darwino, but that's one of many. You could also do worse than learning a mature platform like Ruby on Rails (heck, you could run that on JRuby on Tomcat or WebSphere). Take some time to learn about reverse proxies and modern web-server setups. Basically: something. Just do it with an open mind, and don't balk at the first thing that's more complicated than the Domino equivalent. I think a bit of that will serve you very well.


* And, to be fair, of the overly-conservative nature of enterprise programming.

Other than, of course, being one of the progenitor NoSQL databases.

Code Safety and Pedantry

  • Jun 3, 2016

Lately, I've been musing a lot on the topic of code "correctness" - that is, beyond the normal case of wanting code to do what I intended, and further into the realm of sweating even extremely-miniscule details. A lot of this is due to my continued watching of the evolution of Apple's Swift language (I highly recommend following Erica Sadun's blog for this). Swift is very much in the camp of "make sure all your 'i's are dotted and 't's crossed" languages, as opposed to more fast-and-loose languages like JavaScript or Ruby.

I've gone back and forth on the overarching concepts from time to time. I've long been a big Ruby fan, and a lot of that is because of a general feeling that, if you let go of a lot of the "strict old aunt of a compiler" restrictions, you gain a tremendous amount of expressiveness and productivity with few real-world problems. On the other hand, being immersed in Java all the time has shifted my brain to appreciating the benefits of stronger compile-time checks (at least on paper). Overall, I'm more on the latter side than the former now, double-edged sword though it is. This is why I've been diving into things like aggressive null analysis in my code. Even when it seems like it's being a pedant, there are certain classes of bugs that it finds that I wouldn't even normally think of on the fly. For example, the null checker flags this as being a potential NPE:

if(this.foo != null) {
	this.foo.doSomething();
}

My first reaction upon seeing that was along the lines of "you're full of crap, Eclipse", but then I noticed the small path a bug could take to creep in: multithreading. If I'm in a situation where the object containing foo is used across threads, there's a possibility where Thread A would evaluate this.foo != null to true and start to step in to the block. Then, Thread B would get its turn on the processor and evaluate this.foo = null in another method. Thread A would then pick up and try to call doSomething() on the newly-minted null. So I've swallowed my pride and started writing safer code like:

SomeObject localFoo = this.foo;
if(localFoo != null) {
	localFoo.doSomething();
}

What I've always admired about Swift is that it takes these sorts of lessons to heart and adapts the syntax to suit. My interest in code-correctness pedantry in Java has led me to write out verbose abominations like this:

private final @NotNull String foo;

Three of the conceptual tokens there are purely to say things that are best practices to start with: I don't want this property accessible outside the class, I want to make sure it's assigned during construction and not change thereafter, and I want to ensure it's not null. The Swift variant is:

let foo: String

Same thing, half the typing. And, as a bonus, since nullability checking is built in to the language and not a by-convention thing like the null annotations in Java, I can be sure that the rules will be applied. That sort of thing is the dream! But, since Java is the best language for the work I'm doing for now, the important thing is that it at least suits, verbose or not. In a lot of languages, it gets much more difficult to have this sort of assurance.

So, hassle as it is, I suggest that other Domino developers, on their paths through Java, consider picking up the same habits. For every time you run into something like Java complaining that it can't convert a List<String> into a List<Object>, diving fully into null checks and immutability will save you a late-night crash report and angry user. As you develop more in Java, give it a try.

Darwino for Domino: Conceptual Overlap and Distinctions

  • Jun 1, 2016

I've talked a bit so far about how Darwino related to Domino from a development perspective, but I think it'd also be useful to delve into specifically which concerns the two platforms address, to see where they overlap and where they don't.

There are two main categories to cover, since Darwino inherits Domino's unusual trait of pouring over from "database" to "app-dev platform".

Database

As I covered a few posts ago, the two are similar at a conceptual level, both being replicating document databases with document-level access control. Aside from the difference between an NSF note's format and a JSON document, the main distinction is that Darwino doesn't cover the actual physical storage of data. Instead, it is based on top of existing SQL servers of various stripes (PostgreSQL and SQLite being the most common). A Darwino application creates a series of tables and uses them as the backing store for the conceptual document database.

This has a number of implications. The main one is that there isn't a "Darwino server" as such - instead, there are SQL databases and Darwino applications acting in tandem. In developing an application, this isn't generally a concern: the Darwino APIs are the same across each database, in the same sort of way that a Domino application doesn't care about the ODS version. However, being backed by a SQL server has some distinct advantages: the server can be administered and optimized using the same knowledge you would use for a "normal" SQL-backed app, and the ability of modern DBs to index JSON data opens up a world of possibilities (think NSFDB2, except good).

The flip side of this bleeds into the second category, as it means that a Darwino application consists of at least two parts: the SQL database and the application, which veers from Domino's "everything in one package" promise slightly.

Application

Things diverge most significantly (though at least as promisingly) when it comes to the application level. Domino has a few "official" ways to develop applications (Notes, legacy web, and XPages) and then hooks to sort of act like a Java EE server, albeit with some notable limitations. Darwino, on the other hand, exists as a sort of "glue layer" in between the database and the application: lower-level than XPages but higher-level than just a database driver.

Darwino provides a common platform for writing Java-based applications, with various services for managed beans, user directories, and so forth, written to work consistently across all of the platforms it targets. Again, this starts out similar to Domino, but diverges in the areas where Darwino takes advantage of other technologies.

At the low level, since Darwino's main requirement is "a Java runtime", it is able to run smoothly on various Java EE servers, on Android, on iOS, and on pretty much anything that provides a capable-enough Java environment (such as, say, Domino). It also, incidentally, means that it works great on Java 8.

At the high level, Darwino doesn't prescribe a specific UI framework, so the field is open to use any of the tremendous array of rapidly-developing Java frameworks on the web side and, as desired, native UI toolkits on mobile. There's a bit of an inherent bias towards REST+client JS applications, since then the same code entirely can be used on both web and mobile (as not every Java web tooklit works on the mobile mini web server as it is now), but that's not obligatory.

Overall

So the overall idea is that Darwino doesn't solve every problem that Domino does, but the problems it chooses to farm out are in the areas where that brings tremendous benefit. In each area where Darwino uses third-party support, it benefits from the tremendous advancements made in recent years, without requiring jumping through weird hoops to get modern techniques to work.