[neilv]

Their very first example question is "Do you use source control?", taken from a year 2000 post of Joel on Software.

I think this is no longer a good question. (The answer will almost certainly be the affirmative, and the person asking it might sound unaware.)

[reubeniv]

You'd think that, but some legacy systems might not have it, I don't want to name and shame, but a lot of our code is on an IBM Series i (used to be called as400?) and the source control for it is basically manually adding a comment with the ticket number next to any changes - unsurprisingly they're a bit coy about the technologies used in interviews and target people fresh out of uni who don't know any better, my recruiter told me it was a Java role lol

[avgDev]

We've got as400 at work, surprisingly a lot of companies still run it. Thankfully, I do 0 work on it and focus on other applications.

It is becoming a super niche market, once the dinosaurs die off, whoever can come in and work on it will make bank. In fact my company is getting pretty anxious about it, as the cost of even migrating from it is going to be high.

[neilv]

I've thought of doing a startup for that kind of problem: basically, we'd partner-ish with one enterprise customer as a kind of skunkworks attempt to migrate their legacy system to some other platform, at low cost, by leveraging the language development and DSL facilities of Racket. We could hire some top engineers because we tell them we will pay them money to hack Racket. And we keep rights to the Racket part. That first solution is biased to the first customer -- the particular dialects/versions of language they use, other software/facilities they use, their programming conventions and internal libraries, and how they want to map to which new platform. The next enterprise customer we go to, we have a lot less tooling work to do. After a few iterations of that, we have either a semi-turnkey solution, or a proven approach for our team of highly-paid consultants.

As a bit of evidence in support of Racket being a secret sauce for this, I'd point to how ITA Software (before they were aquired by Google) leveraged Lisp to integrate a new, modern node with the IBM legacy airline reservation system network. They publicly stated that a Lisp was what made this effort viable.

(But doing a Web site or app is so much easier and less risky. VCs are set up to give a site/app dotcom wads of money, and want to see you go through the funding rounds and acquisition/IP. And the technical problems are usually well-understood from the start, and it's just a matter of execution. And you can pick a site/app idea that doesn't involve having to do difficult enterprise sales courtships. Also, personally, given that I started working young, so some of the dates on my resume cause my job applications to be deleted instantly, I'm not anxious to be adding showstopper keywords like AS/400 to my resume.)

[arethuza]

I don't think it is unreasonable to ask something like "What's your strategy for source control?" - not everyone uses git for everything and even if they do there is plenty to talk about.

[neilv]

Good point; the question could be, "What source control do you use?" Or ask about unit testing, CI, deployment, etc.

[arethuza]

If asking something like that you may want to carefully ask how widely adopted the policies they mentioned actually are....

[smudgymcscmudge]

“Yes. We do have a unit test. An intern wrote it last summer.”

[arethuza]

"We think he ran it, but he left" ;-)