top of page
Australian industrial software & local integration
1800 560 854 info@prosightsoftware.com


TLS Works by Name but Fails by IP? Match the Certificate Identity
A monitoring endpoint connects successfully by its DNS name, but the same client rejects a connection made to its numerical IP address. Reaching the same server does not make those two names interchangeable for TLS identity checking.
Nov 133 min read


An MQTT # Subscription Can Miss $ Topics: Test the Filter Boundary
Diagnose MQTT 3.1.1 wildcard filtering without confusing topic matching, subscription permission and publisher activity. Use a bounded commissioning test.
Nov 134 min read


Clearing an MQTT Retained Message Is Not the Same as Publishing Zero
Distinguish zero-byte retained-message removal from a zero reading. Test fresh subscriptions, existing consumers and stale state before an industrial MQTT handover.
Nov 124 min read


CSV Notes Span Two Lines: Count Parsed Records Before Splitting a Handover File
An export contains 100 maintenance records, but a line-based check reports more rows and a file splitter creates broken fragments. A quoted note can contain its own line break. Parse the agreed CSV dialect before counting records, assigning error locations or dividing the file into batches.
Nov 114 min read


MQTT Connections Keep Alternating? Check for Duplicate Client Identifiers
Two gateways can each connect successfully when tested alone yet repeatedly disconnect each other when both run. In MQTT 3.1.1, a reused client identifier is one cause worth checking before changing network timeouts. Compare the identifiers and connection timeline while preserving the intended session behaviour.
Nov 104 min read


Sensor Arrays Change Order: Map Measurements by Identity, Not Position
Prevent changing array order from swapping measurements. Define channel identity, reject ambiguous records and verify a controlled synthetic test.
Nov 93 min read


Timestamp Offset and Zone Disagree: Preserve the Conflict Before Import
Preserve conflicting timestamp offsets and named zones before import. Define a project policy and verify representative records against current zone rules.
Nov 93 min read


Negative Zero in Telemetry: Keep Direction Separate from a Rounded Display
Trace signed zero through parsing, storage and display. Define physical direction separately from a rounded label and test the deployed data path.
Nov 93 min read


Micro Sign or Greek Mu? Keep Unit Labels Separate from Measurement Identity
Two unit labels can look identical while containing different characters. Diagnose the actual character sequence before renaming fields, and keep engineering
Nov 83 min read


CSV Decimal Commas: Preserve Field Boundaries Before Converting Numbers
A decimal comma can be valid numeric notation and still break a comma-delimited handover. Parse the file’s structure first, then apply the agreed numeric con
Nov 83 min read


HTTP 202 on a Telemetry Import: Track Completion Before Declaring Success
An accepted upload is not necessarily a completed import. Keep submission, processing and reconciled records as distinct states so a monitoring handover cann
Nov 83 min read


CSV Header Looks Correct but Will Not Map? Check the UTF-8 File Boundary
An invisible UTF-8 signature can become part of a CSV header in an unsuitable import path. Inspect bytes and decoding before renaming fields or dropping characters.
Nov 74 min read


Averaging Site p95 Values? Rebuild the Combined Latency Distribution
Site p95 latency values cannot generally be averaged into an overall p95. Define the observation population, percentile method and retained distribution evidence.
Nov 74 min read


Combining kWh per Unit: Divide the Totals Before Ranking Production Batches
Averaging batch kWh-per-unit figures can misstate overall energy intensity. Use matched energy and production totals, with explicit boundaries and missing-data rules.
Nov 74 min read


NaN Is Not a Sensor Reading: Validate Finite Telemetry Before Charting
Separate invalid numeric results from real zero and valid measurements at the integration boundary.
Nov 64 min read


JSON Configuration Updates: Keep Null, Missing and Zero Distinct
A configuration update that includes null may remove a setting rather than mark it unknown.
Nov 53 min read


A Fresh HTTP Response Can Contain Old Sensor Data: Test Both Ages
A dashboard can receive a successful HTTP response while displaying a measurement captured long before the request.
Nov 53 min read


Temperature Differences in Dashboards: Convert the Interval Without the Offset
Separate temperature readings from temperature differences when mapping industrial data. Worked Celsius, kelvin and Fahrenheit examples and acceptance checks.
Nov 44 min read


Irregular Sensor Samples: When a Simple Average Answers the Wrong Question
Compare sample and time-weighted means using an industrial load example, with explicit interval assumptions, gap handling and a practical acceptance checklist.
Nov 34 min read


Industrial Daily Reports Across Daylight Saving: Define the Day Before Comparing Totals
Define local-day and shift report boundaries across daylight saving, retain UTC evidence, and compare energy totals using the actual elapsed duration.
Nov 34 min read
bottom of page
.png)