Welcome to The Long View—where we peruse the news of the week and strip it to the essentials. Let’s work out what really matters.
This week: Translating legacy COBOL code to a slightly more modern language, and Arm will go public (again) next month.
1. IBM Code Assistant: COBOL2J2EE
First up this week: Crufty old mainframe code getting you down? Watson to the rescue! That’s the theory, anyway.
Analysis: Not as simple as it sounds
Automated AI translation of undocumented COBOL business logic? The Devil’s in the detail.
Kyle Wiggers: IBM taps AI to translate COBOL code to Java
“Large context window”
COBOL, or Common Business Oriented Language, is one of the oldest programming languages in use, dating back to around 1959. It’s had surprising staying power: … There’s over 800 billion lines of COBOL in use on production systems. … But COBOL has a reputation for being a tough-to-navigate, inefficient language.
…
Looking to present a new solution to the problem of modernizing COBOL apps, IBM … unveiled Code Assistant for IBM Z, which uses a code-generating AI model to translate COBOL code into Java. [It] is designed to assist businesses in refactoring their mainframe apps, ideally while preserving performance and security.
…
Code Assistant is powered by a code-generating model, CodeNet, that can understand not only COBOL and Java but also around 80 different programming languages. [It] was trained with 1.5 trillion tokens … has 20 billion parameters [and] was engineered with a large context window — 32,000 tokens — to capture the broader context.
Need analyst reax. Andy Thurai obliges:
Code generator LLMs are dime a dozen in the market now. … However, the code spewed out by these solutions can be security risks and infringe on IP and proprietary software.
…
One of the major issues with mainframe modernization is finding skilled developers. … Finding developers who can understand COBOL and program in Java is even more difficult. … Code Assistant could be useful to fill these gaps. … Let’s hope IBM can execute.
What can COBOL do? Call on Carly Campbell:
When I get this question, I ask, When’s the last time you:
– booked a flight? ✈️
– moved money from one bank account to another? 💸
– paid a bill? 💳
– withdrew cash from an ATM?
…
Imagine how long some of those applications have been around. They weren’t built on the cloud. … They’ve been in use/development for 30+ years.
Will it work? Steven Sinofsky is not hopeful:
IBM has been trying to convert COBOL to a current language for generations. Maybe this will work but not likely.
Neither is pipo234:
COBOL’s code base is valuable because it works, and expensive because no one dares making changes. … The hard part is stasis and the reason isn’t COBOL, but the legacy context.
…
For years, software has been migrated away from COBOL. Starting afresh, re-implementing, manual translation, transpiling. What remains, is there because it works and no one want to touch it.
…
It is usually not the tedious translation task but the lack of automated tests, foreign function interface calls, archaic hardware quirks, event orders and real-time constraints that are causing the issues. Not sure an LLM is very helpful here.
COBOL? GIGO, more like, thinks silentbozo:
Will it magically figure out what code is actually valid? Figure out what the business rules are that everybody relies on but nobody knows? A lot of this old COBOL code is a pain to work with not because it is COBOL, but because it is old legacy code with little … documentation.
…
**** that is properly formatted and using standard modern patterns is still ****.
DJV brings the ob. Douglas Adams reference:
A case of “accounting software in” and “Vogon poetry out.”
2. Arm Will Go Public (Again) Next Month
SoftBank is letting go of the reins, finally. The owners of the world’s most successful instruction set is selling it back to the stock market.
Analysis: Keep watching
Why should DevOps care? As I’ve said before, ARM chips are an increasing fixture in cloud and on-premises data centers—especially those that value “performance per Watt.”
Milana Vinn, Anirban Sen, Stephen Nellis and Max Cherney: Arm’s clients turn IPO into tug of war
“Switzerland of chips”
Customers of Arm that have held talks about taking a piece of the IPO include Apple, Amazon.com, Intel, Nvidia, Alphabet, Microsoft, Samsung Electronics and TSMC. … Arm is hoping for a valuation of up to $70 billion in the IPO, which will launch on the Nasdaq next month.
…
None of these companies’ investments in Arm’s IPO are certain. … It is possible that some companies will balk at the price expectations.
…
The details of the IPO discussions … illustrate how the company’s neutral status as “the Switzerland of chips” remains a flash point. … Arm’s customers view its semiconductor designs as an indispensable resource. They are used [in] over 30 billion chips annually, powering 99% of the world’s smartphones and everything from the tiniest of sensors to the most powerful supercomputers.
How much? senttoschool does the math:
Nvidia offered to buy ARM for $40B. Right now, Softbank wants a $60B – $70B valuation when many tech stocks have lost 90%+. … What justifies such crazy increase?
…
No doubt Softbank will try to sell ARM as an AI company. It’s not. … Their biggest customers are also their biggest competitors. This puts a cap on how much profit they can make.
What’s the secret to ARM’s success? Ask bazza:
Acorn … set up ARM to generate more value. … They have simply been good at being incredibly useful whilst not taking the piss on licence costs.
The Moral of the Story:
Life’s tough—but it’s tougher when you’re stupid
—John Wayne
You have been reading The Long View by Richi Jennings. You can contact him at @RiCHi, @richij or tlv@richi.uk.
Image: Kevin Wang (via Unsplash; leveled and cropped)
