February 15, 2013
3000 pro uses open source version control
We've been polling the 3000 community about its choices for development tools, but the range runs wider than QUAD or versions of Notepad. One enterprising veteran has tapped the free, open source toolset git to create a batch transfer system for EDI.
The git solution is one of those software choices that seems to defy the traditional structures for care and feeding of software. Like the Joomla Content Management System, git is supported by a vast range of users, comes free of charge for any Windows, Unix or Linux-based workstation or server, and is used by very large companies as well as untold thousands of smaller ones.
One 3000 IT pro, James Byrne of the trading specialist and freight forwarder Harte & Lyne Ltd., checked in to report how git is helping him manage the development of new modules which connect to newer enterprise environments. The git techology supports Behavior Driven Developments. BDD provides developers and business analysts with shared tools and a shared process to collaborate on software development.
Last year I had to create an EDI batch transfer system from one of our suppliers into our billing system hosted on the HP 3000 and written in PowerHouse. For that project I created a git repository for the HP on our source archives' Linux host, and then transferred over all of our source code, job files, udc and cmd files -- and anything else I believed to be locally developed source -- into the git repository using the HP 3000s HFS layout.
I then checked out the specific directories and files into a working directory on my Linux workstation, wrote the new stuff and edited the old stuff in GVim, and checked everything back into the remote repository.
Byrne said he then FTP’ed the new stuff onto the HP 3000 and ran it. "If there were any bugs -- and when are there not? -- I edited the source on the workstation, checked it in to the repository, and FTP transferred it from there to the HP 3000 for the next iteration."
It seeems to me that written out it appears more cumbersome than it actually is. It all went fairly smoothly once most of the gotchas and ‘oops-I didn’t-know-that’ were gradually uncovered and weeded out the the workflow.
One of the major benefits of doing things this way was that everything was built using BDD methodology and the new systems is covered by reproducable tests. Recently a change occured external to our system that broke one of the transfer scripts. We were able to identify the exact problem in our code and fix it with remarkably little effort in an amazingly short time, all because the test suite identified exactly where the exception was occuring and in what way the new behaviour varied from what was expected.
Byrne said the next thing he expects to be writing for, if not actually on, the HP 3000 is a set of Quiz reports to extract the company's 3000 database data into XML files, for transfer and loading into a new billing system. "After that is done," he said, "it seems very likely that then we will bid adieu to our old workhorse."
No more trying to figure out what runs on
MPE/iX or where to find it. No more worrying
about availability! www.MPE-OpenSource.org
is all things MPE/iX: Open Source packages,
freeware, scripting, plus loads of tools
and information to keep your 3000 system
alive and thriving!
The comments to this entry are closed.