Thank you for your insight and feedback.

The “Mainframes” proposal has been cleared to continue.

original post:

There is a new Mainframes proposal in Area 51 about to undergo a final evaluation to see if it should continue efforts towards becoming a full site:

Proposed Q&A site for people who program, operate and want to learn more about mainframe and mainframe-specific technologies

Please have a look at this proposal and the top example questions. We don't generally split off a subject simply to give a sub-topic its own space, but the audience proposing this site has curiously little participation/overlap with our existing communities.

I'm trying to determine if the proposal audience is largely unaware of sites like Serverfault (and related tech/programming sites) or if they feel these topics might not be a good fit on our other sites (in either subject breadth or culturally). I'm trying to determine the suitability of this subject for sites like Serverfault.

Are the example questions predominantly on topic (and welcome) on Serverfault and related sites? Or should we tap this impressive effort (~2-3 weeks to complete Definition) to create a new community?

Any feedback welcome below. Thank you.

  • 1
    Putting the question of whether it should be a separate site to the side, that proposal definitely needs lots more eyes on it. I just used up all of today's close votes and almost all my downvotes on it. (Not implying all my votes are right, I'm not very familiar with the topic, just looking for things which don't seem to fit the SE format.) There's a huge number of broad/opinion questions. – curiousdannii Jun 14 at 7:05
  • 1
    @curiousdannii Why on Earth did you do that? Especially having admitted that you are unfamiliar with the topic. Do you have have the experience to downvote those questions? – Steve Ives Jun 21 at 19:10
  • 1
    @steve It doesn't take any domain experience to recognise shopping list questions and super opinion based questions. The questions I didn't understand I left alone. And see the answers below for other people that also identified many shopping list questions etc. – curiousdannii Jun 22 at 0:02
  • @curiousdannii As is pointed out further down, each site is allowed to set it’s own rules on what is/is not deemed a good question. Why not let this proposal - in which you have no interest at all - go ahead and let those who want it determine the good questions? – Steve Ives Jun 22 at 7:41
  • @SteveIves (1) No SE site allows completely open ended or opinion based questions - closing such questions on Area 51 is completely normal. (2) Area 51 is open to all, as the whole point of the proposal system to put it before as many eyes as possible in order to give each site the best chance it can have to succeed. (3) You would do well to stop assuming things about me. What could possibly have ever made you think I have no interest in the site? – curiousdannii Jun 22 at 8:36
  • Yes, there's a definite spread in the Venn diagram of folks who use mainframes, and those familiar with the Stack* family of sites. Several of the questions I'd agree should not be there... a learning experience for those who are new to the sites and the type and phrasing of questions. (And that's from an IBMer working on mainframe tech, a submitter of a proposal years ago for mainframes, and a Stack* regular.) – Chris Poole Jun 28 at 13:36

This is one of those areas where logical and emotional approaches give different answers.

Logically, much of what mainframes deal with is entirely topical here and we welcome them.

Emotionally, they're radically different parts of the industry. Mainframers are a small minority compared to the Windows/Linux administration population, and like the now-dead NetWare community have a strong sense of identity with their product.

Given that, a site dedicated to mainframes is far more likely to engage the mainframe population than a redirect to here.

  • 3
    That was my initial impression; that this is more about an inherent community [compatibility | incompatibility] rather than scope. – Robert Cartaino Jun 13 at 17:23
  • 2
    I pointed out elsewhere the machines and their OSs are profoundly different from the typical x86 (or POWER, SPARC) server running Unix or Windows. Because of that either the questions that make sense there would make no sense here, or the answers would be completely useless to users of more typical systems. – rbanffy Jun 14 at 15:22
  • 1
    +1, I fully agree. And by the way ... for some reason this remMINDs me about ... MIND ... aka "Mainframe Is Not Dead" ... – Pierre.Vriens Jun 16 at 8:59

Background: Long ago at University I was a mainframe user and spent some time learning about the IBM and compatible systems the university had and a bit about their operating systems. Later (but stll al long time ago) I was a long-time minicomputer user and sysadmin and had some exposure to higher-end minicomputers that were getting close to mainframe capabilities.

I started by giving quick reactions to the first bunch of example questions, but I think that's not what you want. I left the reactions below, just in case they're useful.

Overall, a bunch of the questions seem like "knowledgeable-beginner-looking-for-more-understanding" questions - looking for an overview, some explanation of why things are the way they are.

i.e. not a good fit for StackExchange, the answers require almost a book.

As for why people who are interested in that stuff aren't participating - there just aren't enough enough. Compared to the number of Windows Server or Unix/Linux servers, there just aren't a lot of mainframe systems around, so far fewer mainframe users/admins to (possibly) become interested in ServerFault or SE. Also, since the ecosystem is smaller (lots of IBM stuff, a few other suppliers), I would think most of the expertise is "owned" by IBM and a few vendors. If you want to know about TPF internals, you have to talk to IBM, there are no outside experts...

Here are some quick judgements: I wouldn't necessarily (and couldn't necessarily) defend these judgements, they're just my initial reactions...

In z/OS assembler, what does it mean (and what do I have to do) when I’m told my program needs to be reentrant?

Looks like more of a programming question

What is a dataset, and how is it the same, or different, than a file or directory on a *ix (Linux, Unix) system?

This would be considered too basic on many SE sites, the equivalent of "what's an inode in Linux"

I just did an apt install hercules. How do I install an operating system on it?

Seems like too basic and too broad - How do I install Linux? How do I install Windows...

When JES2 complains of a SPOOL shortage, what’s the difference between TGS shortage, JQE shortage and JOE shortage?

Seems ok

I’m having difficulty in a multi-step job Involving GDGs. I create a +1 data set in an early step but in a following step, I am getting “not found”.

On ServerFault, this would fall afoul of our "troubleshooting" close reason.

What DISP parameter will let my batch job build a new loadlib while I have it allocated in TSOLIB?

Seems ok

How can I find out what "RACF group" I'm in?

Don't know, maybe ok, maybe basic?

What are the serialization mechanisms available in z/OS?

Seems like too basic and too broad - asking a broad question about OS fundamentals doesn't work well on SE

Where can I get a free prebuilt operating system for an emulated IBM mainframe?

I want to access MVS data sets using REST APIs ... how can I do that?

Both of these seem like shopping questions

When would I use JCL and when would I use the Unix System Services (USS) shell to perform a task?

Maybe ok, but might be too opinion-based, like the difference between unix shells.

I need to allocate a dataset. What does allocate mean, and what are record formats, logical record length, and number of tracks?

Another broad question about basics

How do I access data sets from the USS shell? How can I create a Certificate Signing Request to install new certificates for web sites in a Sysplex?

Maybe ok

Can I use my current open source tool chain for z/OS development? (It currently has Git, Jenkins, Ansible as the primary capabilities)

Too broad

What are the fundamental differences between JES3 and JES2 ?

Maybe too broad, maybe ok... maybe there are a few really significant differences.

Why is the FORCE command dangerous? Is it safer to use the similar functions provided in monitoring software etc?

Probably ok

  • All these comments would be better if made in form of suggestions on how to improve the actual examples. This is an exploratory phase and such input is very valuable. – rbanffy Jun 22 at 11:17

There are a lot of terms and phrases there I've been happy to forget since moving on from mainframe ops.

Without commenting on the specific questions, in theory mainframe questions in general are welcome on server fault (iirc I answered at least one way back) but they're tightly focused to a small audience and as such I think they're probably better off with their own mainframe specific site... though I do wonder if they'll have enough throughput to make it reach the activity threshold.

  • 1
    I agree, and yes I still seen sale on mainframe lately, people think it’s dead but there is still some niche that still use those, but sooo limited that I agree with you that it’s a target for really small audience. – yagmoth555 Jun 16 at 11:31
  • The issue we had in the past—we tried to get a site to beta years ago—is that there's very little overlap between the mainframe experts, and existing Stack* users. And you need that to get it to beta in the first place.If we can succeed in doing that, I think the site will do well... there are a lot of new folks on the platform that pretty much expect a Stack* site or equivalent, since that's what what they used at college etc. – Chris Poole Jun 28 at 13:49

Are the example questions predominantly on topic (and welcome) on Serverfault

No, I don't think so. Several of them are shopping questions (eg "Where can I find free online IBM Mainframe Manuals and Programming Guides?", "Where can I get a free prebuilt operating system for an emulated IBM mainframe?"), and most of the rest of them are development questions, not system administration questions.

  • 2
    I should always preface questions of scope with "assume the questions are suitable for the Stack Exchange format in general" because I am asking about scope and topicality... not form or composition. Sorry about the confusion. – Robert Cartaino Jun 13 at 0:26
  • 2
    The most frequent question you hear in mainframe circles is "where do I start?". The machines and OSs are profoundly alien to someone who grew up on Unix-like systems. – rbanffy Jun 14 at 15:19

So my question is: are topics that won't fit serverfault (too broad, too basic) combined with inherent community a good reason for an own site?

Or would these questions be automatically "bad, should be closed" there, too?

  • Each site gets to set its own rules within the broad Stack Exchange guidelines, so their own site will work well in that regard. – Rob Moir Jun 16 at 12:05
  • The onboarding process for mainframers is very different from the "normal" crowd. Anyone can buy an x86 and install Linux on it and get going, but buying a mainframe and getting it up and running is so rare it's notable. At the same time, emulators only will take you so far unless you violate the licenses and run software you are supposed to run on real IBM metal. And that is only the IBM space - don't even get started on MCP and its offspring. – rbanffy Jun 22 at 11:23
  • Because of that, I expect a lot of questions about "where is the manual for X" (IBM doesn't help you much there) and "I got X. How do I run Y on it?" which are questions I had myself and some I never adequately answered and just gave up. – rbanffy Jun 22 at 11:23

The proposal has already passed, but I'll give my $0.02 anyway.

The mainframe is quite unique in that it jobs roles are often quite distinct, but solutions cross these boundaries. To solve a problem, some part of it may be for a developer to implement, and another part for a sysprog (sysadmin); a correct answer to a question would really cross from Serverfault to Stack Overflow, or elsewhere. Sometimes a question can relate to software, but since the stack is very controlled, involve dependencies on hardware too... For these reasons, I think it's valuable to have one mainframe site, instead of trying to send folks to three different sites and hoping they find a good solution that they can relay to others for future reference.

The mainframe is also—for historical reasons I think—bugged by forums where the standard answer is "You didn't read the manual properly. Read it again and then come back." A new, consistent site, is a great way to start and reject anything unhelpful like this; it should be great for experts and newcomers alike.

  • I think you underestimate the importance of reading things. I think you overestimate the power of an SE site. Look at devops.se for example - there are likely orders of magnitude more 'devops' than that are mainframe users, they get ~3 questions per day. I wish your site well but won't be at all surprised if it turns out to be ::tumbleweed:: – Iain Jul 1 at 19:38
  • I would assume the questions on this SE would be much more durable than the questions on the devops site. If the pace of change doesn't change significantly in the mainframe programming space, I'd suppose answers on it will be useful for at least a decade without much change. – rbanffy Jul 9 at 7:18

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .