"Port & Mend" over rewrite. Reducing the cost & risk of dealing with legacy code
summary
Porting a VAX FORTRAN control program to a PC (.NET Fortran compiler), running it in parallel to compare outputs, and fixing race conditions yields a working replacement tested against the old system and switched in November after matching outputs; a four‑person team completes the port from spring to November. Migrating parts of a PHP7/Yii1 monolith into PHP8 containers lets one team deliver five modules in four months versus one module in a year. Uses parallel plant validation, incremental porting and refactoring, AI for code mapping and CDK documentation, and produces safer, cheaper, deployable results.
AI-written summary · about
chapters (12)
- 0:00Legacy code challenges
- 1:09Legacy modernization stories
- 7:14Porting FORTRAN to .NET
- 9:44Rescuing PHP monolith
- 13:45Port vs rewrite comparison
- 14:51Kintsugi metaphor
- 16:15Mapping legacy systems
- 18:45Safety in code changes
- 21:17AI for legacy analysis
- 25:14Port and mend for leaders
- 27:21Feasibility of port and mend
- 29:12Summary
related talks
53mPorting Code - Lessons Learned
40mRefactoring Huge Legacy PHP Applications - From 100 hours to minutes
21mTransforming Legacy Systems: A DevOps Journey to Modernization
59mReplatform in a Year or Die: The Tale of a Turnaround
01h 07mRewriting 15-Year-Old Code
55mModernizing Legacy Code - While Not Bringing Down Prod
40mLegacy Evolution – The Innovation Opportunity!
51mLegacy non-x86 modernization with