Common Lisp Implementations in 2023

> Are there studies on the rate of project success comparing effort in A: JDK, .NET; and B: Go, Lisp

No, there aren't and anyone saying otherwise is lying.

Almost all relevant data is tightly locked up in corporate software development silos (FAANG, enterprise middle ware, website or mobile dev shops, game devs, etc). The few studies that we have are limited to academic (basically toy projects) or OSS settings (and we don't really have the behind the scenes info from the corporate sponsors) and their sample sizes are pitiful.

Plus, what even is success? Adoption after 1 year? After 10 years (https://www.joelonsoftware.com/2001/07/21/good-software-take...)?

What we can measure instead is the number of developers, jobs, projects started based on public announcements, etc.

And for that it's no contest. Java/C# >> Go >>>>>>> Lisp.

The reasons are less important than the facts on the ground.

Especially since Common Lisp was first proposed in 1982 (41 years ago), so a really long time ago. However, its rivals aren't spring chickens either, these days. Java was launched in 1995, 27 years ago (early Java devs are close to retirement, already), C# was launched in 2000, 23 years ago, heck, even Go was launched in 2019, that's 13 years ago.

So we're in a reality were we have multiple competing ecosystems that are big and developed and have been around for decades. So the mainstream ones will be major for a long time.

It looks interesting, however this is a bit scary:

> WARNING: This software is BETA quality. I use it as my daily driver, but it is still a little rough around the edges and it may accidentally eat your files.

And for QuickLisp, this is scary:

https://github.com/quicklisp/quicklisp-client/issues/167#iss...

> It would be good to do, but there's no straightforward path to do it. Implementations do not all provide HTTPS support, it's not straightforward to make it from scratch or use HTTPS libraries on all supported platforms.

The fact that the package manager (!!!) isn't able to ensure use of HTTPS libraries as its own dependency on all platforms is... super scary.

SBCL's save-lisp-and-die is my all time favourite function name. It's so unintentionally melodramatic :)

Okay but why would you do that in a large company, you don't need every single developer triple booting. You've got 150 people and 25 of are RHEL, 25 are macos and 100 are windows.

and also the 5k license is a one time fee. over the course of say 5 years its

    5100 [first two years] 
  + (1125 * 3) [next three]
  = 1695 a year [asymptotically approaching 1125]

if you want to target both mobile runtimes

  + (5 * 2000)[the mobile runtimes *are* annual although as with desktop you wouldn't actually need everyone to have both, or either] 
  = 3695 a year [asymptotically approaching 3125]

That's before any volume discount (I dunno lispworks but as a point of comparison Sicstus goes from 2550/seat to 1510/seat for just 20 seats) and unlike say Dyalog, distribution is free so only the devs need licenses not any one else in the org.

I'm not saying it's cheap mind you but even at 3700 a year * 100 devs, if your bumping up against sharp edges in SBCL: 370000 only gets you 1-2 devs to work on those edges.

At the size of a Google adding developers works out to be cheaper than licensing software for all the existing devs. But a lot of companies don't have high 3-4 digits worth of devs such that they can break off a team to work on support technology instead of core internal business apps or products.

No matter how you slice it, Java enterprise would only cost $180/yr which is almost 10x better than your best estimate

Let's say you have 25 devs on a 5-year project. 5 of them are mobile focused, 5 are windows focused, 5 are mac/Linux focused, and 10 are focused on the cloud/server side of the project.

We'll make the (probably bad) assumption that you don't need enterprise licenses for the desktop app portion. We'll further assume that windows devs use only windows and backend devs use only Linux while mobile and mac/Linux devs use both the systems they work on.

    First 2 years (because we get a discount on year 1 maintenance)

    $17,000 for 5 64-bit windows professional
    $51,000 for 10 64-bit linux enterprise
    $37,000 for 5 64-bit mac professional and 5 pairs of mobile for 2 years
    $34,000 for 5 64-bit mac professional and 5 64-bit linux professional

    $139,000 for first 2 years

    Next 3 years (maintenance and mobile annual fees)

    $11,250 for 5 64-bit windows professional
    $33,750 for 10 64-bit linux enterprise
    $41,250 for 5 64-bit mac professional and 5 pairs of mobile for 3 years
    $22,500 for 5 64-bit mac professional and 5 64-bit linux professional

    $108,750 for next 3 years

    $247,750 for project.

Now let's do Java at $180/yr/dev. We'll even throw in IntelliJ Ultimate for 5 years.

    First 2 years
    $9,000  - 25 devs for 2 years of enterprise Java
    $14,975 - 25 devs for year 1 of IntelliJ Ultimate ($599/yr/dev)
    $11,975 - 25 devs for year 2 of IntelliJ Ultimate ($479/yr/dev)

    $35,950 for first 2 years


    next 3 years
    
    $13,500 - 25 devs for 3 years of enterprise Java
    $8,975  - 25 devs for year 3-5 of IntelliJ Ultimate ($359/yr/dev)

    $22,475 for next 3 years

    $58,425 for project

Drop the non-enterprise devs from support reduces Java cost from $22,500 to just $9,000. Dropping subsequent years of the new IntelliJ would cut that cost from $35,925 down to $14,975 reducing our total expenses to just $23,975.

I'd calculate these for LispWorks, but they don't even provide that option at all. If you forego maintenance entirely, you're still stuck with $115,000 the first year and $40,000 for the remaining 4 years of subscription to mobile meaning your absolute minimum here is $155,000 and exceeds even the top-end budget for a Java project. If you turn out to need the Enterprise version for the non-backend devs and want maintenance, that cost goes up to $267,875.

The difference in money is enough to provide every team member with a new $4,000 laptop year 2 and again year 4 OR to increase their yearly bonus by $1600.