El Reg reports
Attachmate, the software shop that headhunted Novell and SUSE Linux, is itself being bought out by Micro Focus International.
The mainframe and COBOL specialist is acquiring Attachmate Group from its parent company Wizard in a deal calculated at $2.3bn before costs.
[...]
Attachmate gives MicroFocus access to SUSE and Novell, business units bought by the company in 2010 for $2.2bn. Novell owned SUSE Linux, which it had bought in 2003 for $210m. Under Attachmate, the two were broken apart.
putting 882 patents in its Linux portfolio up for sale to a consortium backed by Microsoft.
SUSE is chief steward of the SUSE Linux Enterprise Server while Novell has been re-shaped to sell end-point management and collaboration software.
(Score: 4, Interesting) by PizzaRollPlinkett on Wednesday September 17 2014, @02:14PM
MicroFocus was originally the vendor of a non-mainframe COBOL implementation that competed with Fujitsu and maybe a few others. The selling point was to do development offline, to not use expensive mainframe computing cycles for syntax checks and stuff that could be done on PCs. The problem was, if you didn't have a mainframe, you didn't have DB2, CICS, VSAM etc. Compiling COBOL wasn't easy without these dependencies. So MF lashed together some technologies to allow you to simulate those dependencies. Most of this stuff was obviated by falling mainframe prices - no one really cared about doing development offline any longer, after a point.
Because DB2 was a dependency, a vendor called XDC wrote an API-compatible (DSNHLI, if you know what that is) clone of DB2. It was supposed to let you run COBOL programs with embedded SQL, and do any DDL you needed to create objects. So, MicroFocus buys XDC right when IBM releases DB2 UDB for Linux and Windows. Oops. Not a smart move. The "U" in UDB was for the Universal API, so this real DB2 (which IBM began giving away) rendered XDC obsolete. Development started moving to Java in Big Blue land, leaving COBOL as a legacy mainframe language for back-end processing.
MicroFocus became a hot potato in the late 90s/early 00s as it was sold from company to company. At one time, an ODBC driver vendor owned it if I remember right. Eventually, having almost zero value, MicroFocus was spun off into its own company again. Now they seem to be acquiring also-ran companies.
I will say the MF compiler was decent. I used it on Linux for a while to test a product that could be called from COBOL and had to be mixed-language compatible. (Including fun surprises like MF COBOL storing binary words in big-endian format on little-endian architectures like x86 - I have no idea how they did math with those words, but it had to be slow.) The most bizarre thing I ever did with MF was compile a COBOL routine on Windows as a DLL and call it from a Delphi GUI program. I can't remember how all that worked, which means the shock treatments were worth every penny.
MicroFocus is basically a really expensive solution to a problem no one has. COBOL is a back-end language for mainframe business logic, mainly in CICS transactions, but new development is in Java with WebSphere. Your front-end Java crud calls MQ Series (now WebSphere MQ, since IBM needs to rename products every few years) to post a message, and your batch CICS back-end does the business logic processing. No one needs to compile and run COBOL anywhere but the mainframe.
Every few years, I see an article about some company migrating something off the mainframe, and I know before reading it that it's a MicroFocus placed article. I guess they have enough business to survive, which is something to give them credit for.
And, remember, IBM has their own COBOL compiler with a full ANS COBOL parser. They call it "Enterprise COBOL" now. They also have code generators for x64/x86. IBM could very easily put MF out of business overnight by releasing an open-source COBOL compiler. Remember, they had Visual Age COBOL running on OS/2 in the 90s, so they have the technology. Probably wouldn't take them more than a few months to create a clang/LLVM or GCC front-end COBOL compiler. But there's zero demand for it, and not worth the effort. People just don't use COBOL enough off of the mainframe to be worth doing.
Attachmate sells a PC-based mainframe terminal emulator, so I guess MF is covering its bases - if you don't want to move COBOL compilation off your mainframe, they'll sell you a terminal emulator.
IBM supports SUSE with their non-mainframe enterprise stuff (WebSphere, DB2, MQ, etc) so there's a sort of synergy here. Makes sense for MF to have SUSE and support customers who rely on it for their WebSphere. I want to say SUSE is one of the distros that have been compiled to run in IBM VM/CMS as a virtual machine image, but I just can't remember and the shock treatments have erased my memory of where to look it up.
(E-mail me if you want a pizza roll!)
(Score: 0) by Anonymous Coward on Wednesday September 17 2014, @02:34PM
Mod parent up. I have passing familiarity with some of those pieces and he's got some good stuff there
(Score: 2) by VLM on Wednesday September 17 2014, @03:49PM
" Most of this stuff was obviated by falling mainframe prices - no one really cared about doing development offline any longer, after a point."
HERCULES to the rescue! Pizza almost certainly knows the software I'm talking about, others can google.
Sure its illegal to run licensed software products on an unlicensed hardware emulator, but with little more than a wink and a nod everyone does it anyway.
Speaking as a guy who suffered thru MF cobol classes around the turn of the century and worked at a mainframe shop in the early 90s there is no avoiding the fact that java is the new cobol. Hyper verbose, NIH to the max, complicated business logic, nobody but fortune 50 companies wants it that way. Everyone else sees it as a liability. Java is like SAP in that way.
I screwed around a lot with hercules and mvs360 and the turnkey system maybe a decade ago. If it was good enough to run a multinational megacorp in 1965, I'm going to fool around with it in 2005 and see what I can learn from retrocomputing and expanding my brain into new (new to me) architectures and thought patterns. Strange stuff.
I wish in 2014 linux had good free batch queuing as good as we had in the 60s with MVS. Someone will likely implement that someday. Hopefully not into systemd. Torque/PBS/Maui just aren't quite up to that level. Close, though.
(Score: 3, Interesting) by PizzaRollPlinkett on Wednesday September 17 2014, @04:17PM
Hercules has always been a raw nerve for me. If there's a "skills shortage" and qualified mainframe programmers and systems programmers are in short supply, why wouldn't IBM release an educational version of its turnkey MVS system (what do they call it? ADCD?) for people to use with Hercules? I mean, it's not like I could do real data processing on a laptop. But I could learn how to use JCL to compile COBOL programs. You'd have a whole generation of college students getting into mainframes for the first time. What better way to learn systems programming that setting up Hercules? They'd have a whole generation getting excited about the mainframe for the first time, and ease their "skills shortage".
Oh, no. What did I just type? Use JCL to compile COBOL programs?
There's a magic potion that makes you forget about that stuff.
Bleach.
(E-mail me if you want a pizza roll!)
(Score: 2) by VLM on Wednesday September 17 2014, @06:10PM
Well the turnkey is 3rd party uncopyrighted all the way. Yeah the ADCD or whatever its call is a full system from ibm but it's all locked down illegal although in theory you can run it. One minor problem is if you think the MVS turnkey is mysterious and huge and complicated, modern zos is like 1000x bigger so its almost hopeless.
This might be some of their fear, you can release it, but nobody being able to figure it out other than the experts might be more bad press than good. Unlike unix / bsd / linux I don't believe you can self teach on MVS, but thats just my personal belief.
I could figure out how to use fortran on the pdp8 all myself with the manuals, but I had to get help to run cobol on mvs.
(Score: 2) by PizzaRollPlinkett on Wednesday September 17 2014, @06:36PM
The MVS 3.8 turnkey system is so obsolete that if you learned it you wouldn't learn anything you could use on modern z/OS. I know 3.8 doesn't have ISPF (the full-screen stuff - have fun using TSO commands without it), and I can't remember if it has JES2 or the older HASP. Even recently, z/OS is getting so heavily dependent on OMVS (the POSIX subsystem) that you basically have to know UNIX, too. But I said goodbye to all that and never looked back, especially after the shock treatments. It is getting hopeless. No one can grasp it all any longer, so IBM is releasing pre-built systems. (That's one reason I wouldn't get into mainframes if I was young - there's just less work to do these days, causing high unemployment, and you're competing with people who have decades of experience. Will be a long time before that works its way out of the pipeline.)
One thing that UNIX/Linux/FreeBSD/etc can't really teach you is file allocation. On MVS/zOS you have to know ahead of time how big your file will be, and what record format it will use. DEC VMS had a little of that, but almost all modern operating systems have abandoned this concept. Rightfully so. (You can, of course, code your own file formats that use fixed-length records. But you have to code your own utilities, too.) But you haven't used a real operating system until you've had to calculate blocking sizes, record lengths, extents, and so on. That stuff is a whole different world.
And the most important thing ... wait, what's that sound? Oh, a drone just landed. With my same-day delivery package! Let me open it... yes! A 64-ounce bottle of bleach!
(E-mail me if you want a pizza roll!)
(Score: 2) by choose another one on Wednesday September 17 2014, @11:20PM
They are better than surviving, look at their share price over past few years (and note that there have also been some very big special dividends to return even more money to shareholders). For a solution to a problem no one has, they are sure selling a lot of it.
More background: the current top management team are the same guys who turned Morse around before selling it to debt-laden monstrosity 2e2 (who destroyed it and themselves). They moved to MF and executed the same kind of turnaround. I made good money on the Morse turnaround and some more since on MF by watching that management team, but I wish I had bet the house on Morse and re-invested all of it in MF at the times I did - be retired by now, more than comfortably.