Abe Voelker's blog Programming stuff, mainly

Final Ode to OpenEdge ABL Part 3: Parting Advice

4 minute read

Yao Ming laughing at OpenEdge

In part 1 of this series I demonstrated how to write Ruby code against an OpenEdge database using the JDBC driver and DataMapper ORM, and in part 2 I showed how to take advantage of this to rapidly prototype a RESTful Web service / JSON API for the canonical sports2000 OpenEdge database using Ruby and Sinatra.

For the final post in this series, I promised to offer some advice to Progress Software Corporation (PSC) on how to advance the OpenEdge ecosystem.

In the spirit of the words “let the dead bury the dead” I’m going to keep this short as I’m skeptical that this post stands much chance of effecting any changes.

The Golden Goose

What PSC needs to do is to decide which of the nasty siamese twins is their golden goose - OpenEdge the database, or ABL the programming language. I’m assuming it’s the database because ABL mainly exists to query it. My advice is to focus on fluffing up the goose, and give the axe to the ugly duckling (or gosling, I guess).

Improve OpenEdge SQL

First, the database is not nearly powerful enough to warrant requiring its own grammar. I would drastically improve support for the lingua franca of relational databases, SQL, by incrementing the SQL engine to a newer SQL standard. OE features that don’t fit into a recent standard can go into vendor-specific SQL grammar (like everyone else does). This would make adding support for OpenEdge databases to popular programming languages very easy.

Deprecate ABL

Trying to turn ABL into a full-featured programming language was and continues to be a mistake on the part of PSC. They simply can’t iterate fast enough to keep ABL up to date with the pace of innovation in the rest of the software industry - and that’s understandable.

To put it bluntly I think ABL should be deprecated. The CLR bridge is the smartest thing that PSC did to the language, and instead of continually trying to play catch-up with developing their own language, they should embrace language openness by doing a few things:

  • Support Mono as a .NET development target, which would allow writing .NET code in the AVM for Unix systems. This would allow programmers to use a much stronger language like C# for general purpose development. This is already a viable option for Windows development, but not currently a general-purpose solution due to the lack of Mono support.
  • Open source the ABL Virtual Machine (AVM). This would give the ultimate in extensibility to ABL coders by allowing them to code in any language as long as it compiles down to AVM bytecode.
  • Open source the ABL language and compiler tools (predicated on last point due to AVM bytecode dependency). This would allow the community to fix bugs and add features to the ABL language without having to wait for PSC.

Marketing Reboot

Drop the “business language” bullshit. Such a concept doesn’t exist. If it did, the end users would be writing the code themselves. Same goes for the silly “4GL” label of false superiority. Coupling a programming language to a database doesn’t make the language look more advanced, it makes it look ridiculous.

The big push on marketing OpenEdge for “the cloud” is a joke. “Cloud” apps are all about services / APIs, which OpenEdge fails at; it doesn’t even have REST support yet. The only Web service it supports is SOAP which is disgustingly ancient and involves a huge amount of glue code, such as generating WSDLs and using the ProxyGen to prepare more Java or .NET interface glue code. If your process is this obsolete and complicated then I don’t think you’re “cloud”-ready.

Also, multi-tenancy isn’t a new feature to most databases and not really much to brag about - I know for a fact that Postgres has supported them for many years (they call them “schemas”). I also wouldn’t consider multi-tenancy to be any kind of requirement for writing a “cloud” app - in fact I’ve seen convincing arguments against using it for such services. So hold off on the champagne.

Quite simply, PSC has historically spent way too much on marketing and too little on actual engineering… that needs to change drastically.

Conclusion

This series has been cathartic. I feel like I have some closure on my past now, and can completely let go of the mental baggage I’ve been carrying of my depraved interactions with OpenEdge.

I plan to continue evolving as a software developer by moving onto more functional pastures, focusing on Clojure at the moment. So far it seems a very good intersection of the “pureness” that I loved about learning LISP (Scheme) in school but with actual practical capabilities rather than being so utterly academic. Of course Ruby will maintain a soft place in my heart, and continue to pay my bills. Stay tuned!