Compare 3270 screen flows before a mainframe migration¶
3270Web can compare exported screen maps from two environments to identify field and navigation differences. Use it as one input to a migration review, then replay representative transactions and verify their business outcomes. A matching screen map is not proof that data, timing or every application path is equivalent.
Keep the comparison reproducible¶
Choose an approved test account and a known initial state on each host. Match terminal model, screen size, code page and the permissions of the accounts. Record the application build, configuration and test-data version with each export. A different account audience can change reachable screens even when the application code is unchanged.
Start with a small journey: sign in, navigate to one enquiry, inspect the result and return. A bundled sample is a useful place to learn the controls before exploring a business application.
Capture baseline and candidate maps¶
- Connect to the baseline test host and use Chaos Mode to explore the approved paths.
- Export the mind map and label it with the host/build/test-data revision.
- Connect to the candidate host, restore the same initial state and repeat the exploration with comparable settings.
- Keep both exports and the discovery reports together. Record excluded or unexplored paths explicitly; absence from an export is not necessarily absence from the application.
See Chaos Mind-Map Compare for export commands and the
baseline/candidate request format. The comparison operates on the two
supplied JSON documents; it does not silently run a new exploration.
Interpret deltas as questions to investigate¶
| Report result | Follow-up check |
|---|---|
| Only in baseline | Was the screen removed, unreachable with this account, or simply not explored? |
| Only in candidate | Is this an intended new path or an unexpected transition? |
| Field added/removed/modified | Does an input moved or resized here affect the recorded workflow? |
| Transition removed/redirected | Does the same PF/Enter action still reach the expected business step? |
| No deltas | Was meaningful coverage achieved, and are business results also equivalent? |
For example, a removed PF3 transition is a reason to replay that navigation step manually and in a recorded workflow. A new field is a reason to inspect its position and semantics, not to immediately rewrite every test. The comparison reference describes how areas are matched and the exact counters returned.
Replay the business result¶
Debug the recorded workflow on each host. Check the expected fields and final outcome, and compare approved screen snapshots where useful. Account for legitimate differences such as generated IDs or timestamps rather than treating every changed character as a defect.
Use Host Compatibility Profiler to compare negotiated terminal capabilities separately. When functional checks pass, run an approved workload through 3270Connect load testing to investigate timing and concurrency under a defined load profile.
Keep an evidence record¶
Retain the two maps, discovery settings, inspected deltas, recorded workflow, assertion results and the known coverage gaps. Restrict access to exports and screens that contain business data. Repeat the same review after a fix; use the resulting evidence to support the migration decision alongside data, security and operational testing.