5 ms·
> In summary we've seen that: Haskell decreases development time... Have we actually seen that or have you just asserted that? Is this really true, and if it i
by pron 10y ago
> In summary we've seen that: Haskell decreases development time...
Have we actually seen that or have you just asserted that? Is this really true, and if it is, by how much? Haskell has been around for a couple of decades now, and has had least two hype cycles (I remember that when I was in university in the late '90s, Haskell was the next big thing). It does not seem to expand significantly even within organizations that have tried it (and that's a very negative signal), with at least one notable case where the language has been abandoned by a company that was among the flagship adopters.
In general, we know that often linguistic abstractions that seem like a good idea in theory -- or even seem to work nicely in small programs -- don't end up having a significant effect on the bottom line when larger software is concerned. People say that scientific evidence of actual contribution is hard to collect, but we don't even have well-researched anecdotes. Not only do we not have strong evidence in favor of this hypothesis, but there aren't even promising hints. All we do have is people who really like Haskell based on its aesthetics and really wish that the the nice theoretical arguments translated to significant bottom-line gains.
This blog post by Dan Ghica, a PL researcher, really addresses this point: there is nothing to suggest that aesthetically nice theory translates to actual software development gains, and wishful thinking (or personal affinity) simply cannot replace gathering of data: http://danghica.blogspot.com/2016/09/what-else-are-we-getting-wrong.html http://danghica.blogspot.com/2016/09/what-else-are-we-gettin...
- radicality 10y agoNot the author and I don't know much besides a little dabbling in Haskell, but I'm sure that quote holds true, but only if the engineer already is an intermediate-advanced Haskeller, which probably translates to at least a couple years of daily Haskell work.
- pron 10y ago> but I'm sure that quote holds true How can you be sure? Have you seen Haskell shops churn out software significantly faster than others?
- willtim 10y agoWe use Haskell for "rapid application development" in the front office at Standard Chartered bank. We have excellent typed APIs over most of the banks systems and can assemble applications in hours or days, rather than weeks or months as is typical. The language is as succinct as python, but thanks to types, purity and many standardised abstractions, we get better code reuse and maintenance.
- pron 10y agoHave you written a well-researched case study? If not, please do. It's very hard to assess such statements otherwise. After at least six years of Haskell at SC, is it spreading to larger software in the bank? If so, how fast? If not, why not? Are other banks really spending orders of magnitude more on that kind of software? I've heard that SC adopted Haskell when Lennart Augustsson -- who wrote the first Haskell compiler -- moved to SC from Credit Suisse. When he left CS, they stopped using Haskell. If Haskell gives an order-of-magnitude difference in productivity as you claim, why did CS drop it after Augustsson had left?
- willtim 10y agoAFAIK, Credit Suisse only used Haskell for one particular project and that project came to a natural end. CS did end up adopting functional programming and F# became their strategic choice. I never claimed "orders of magnitude" productivity from a language alone. Haskell has every weakly-typed mainstream imperative language as a valid subset. You need good people, teams and processes to get the best out of the language. This means it is difficult to do case studies without getting dismissed as anecdotal. Most other banks have settled on Python for similar use cases. I hear on the grapevine that they have significant maintenance problems (unsurprisingly). Haskell is spreading to larger software in the bank and we are currently hiring.
- sheepmullet 10y ago> and can assemble applications in hours or days, rather than weeks or months as is typical. So you are building in hours and days applications that other banks are spending weeks and months on. I'd love to see a few case studies!!
- psibi 10y agoIt's hard to measure these types of things. But if you are interested, there is a related paper regarding that: http://haskell.cs.yale.edu/wp-content/uploads/2011/03/HaskellVsAda-NSWC.pdf http://haskell.cs.yale.edu/wp-content/uploads/2011/03/Haskel... Although the study in the paper isn't very practical, it's still an interesting experiment.
- pron 10y ago> It's hard to measure these types of things. I'll settle for well-researched case studies. > there is a related paper regarding that That's a step in the right direction, but the paper doesn't discuss software development, but prototyping. We know that "theoretically aesthetic" languages do well in specification and prototyping.
- Tehnix 10y ago>I'll settle for well-researched case studies Don't know if it helps, but they have some case studies here https://www.fpcomplete.com/case-studies https://www.fpcomplete.com/case-studies.
- pron 10y agoThat would have helped a lot if those really were case studies. Unfortunately, they're just marketing material for FP Complete (with statements like "In Haskell, Acme Inc. found the perfect solution!"). There's nothing wrong with marketing material, but that's not what I meant by case studies (I meant actual technical reports).
- Tehnix 10y ago>Have we actually seen that or have you just asserted that I'd say they have probably seen it. The author of the post himself works at FPComplete, and they use Haskell for their consulting business. That and there are some case studies on their site (I'm fairly sure they all have experience in other languages too).
- pron 10y agoThere aren't any case studies. Under the headline "case studies" they only have some marketing material.
- willtim 10y agoYou appear to be misrepresenting the blog post. His example of an aesthetically motivated language is Python, not Haskell. Haskell's design was not motivated by aesthetics or superficial ease-of-use, which could well be its downfall. Haskell was and still is motivated by enabling software development gains, e.g. abstraction, composition, concurrency. You may disagree that it has achieved these goals, but to that I would say don't knock it until you have tried it.
- pron 10y ago> Haskell was and still is motivated by enabling software development gains The same can be said about nearly every language. Haskell is first and foremost a research language motivated by answering a research questions, which is what it is like to program using a certain mathematical formalism that is aesthetically appealing to some. Its designers didn't set out to find the best abstractions etc., but to test the validity of a specific theoretical approach. > but to that I would say don't knock it until you have tried it. I'm not knocking it. I'm saying that given how expensive it is to try, perhaps some actual information rather than marketing would do a better job in persuading people to use it. Saying that I'm not persuaded by virtually zero information is no knocking anything, except perhaps the advocacy effort.
- willtim 10y ago>> Haskell was and still is motivated by enabling software development gains > The same can be said about nearly every language. This clearly isn't true. For example many languages optimise for ease of use, shallow learning curve, programming in the small etc. Some optimise for efficiency and runtime determinism. Features like type-systems are great for software development (programming in the large), but do negatively impact ease-of-use. > Haskell is first and foremost a research language motivated by answering a research questions This isn't true. The Haskell committee contains many people from industry. Haskell just happens to have come from academia and is still heavily used in academia. > Saying that I'm not persuaded by virtually zero information is no knocking anything That's a huge exaggeration to suggest there is zero information. There are many experience reports published all the time at conferences and in academic journals, going back decades. None however will give you the quantitative metrics you seem to be asking for, because it is almost impossible to do such a comparison.