📝 DMS-NX_Chapter2_Master_Post.mdv4.4 · 2026-09-05

DMS/NX — Chapter 2

The forum's search and document system, formerly known as ISE.


What's this?

DMS/NX (Document Management System / NeXt generation) is the new name for ISE, the search engine built for this forum. This is a rebrand only. Nothing has been replaced or removed. It's the same engine and the same indexes, and your searches behave the same way.

This post is the new Master Post for the project. It replaces the old ISE Master Post (Chapter 1) as the place to find:

The Chapter 1 thread stays up as the project's history.


What DMS/NX searches

DMS/NX has five search sources. Each one is tuned for a different kind of content.

Source Searches Notes
ISE (Posts) Forum post text The original engine. Word-index lookup, then ranking.
PDF Text inside PDF attachments Finds documents, not just the posts that link them.
Misc Text inside flat-file attachments Any plain-text format: .json, .txt, .py and so on. Excludes .md, .html and other formats handled elsewhere.
ISEmega Posts + PDFs + Misc together One combined, ranked list. Returns up to 180 results by default.
ISEmedia Images and video AI (CLIP) image matching. Works differently from the others. See below.

Why ISEmedia is different

The four text sources start by looking your words up in an index, so every result actually contains your terms before ranking starts. AI image matching has no index like that. It scores every item for similarity, so without a filter it would return near-misses.

To keep results relevant, ISEmedia first runs a keyword gate: an item must literally match your search terms before any sort mode is applied. ISEmedia also returns at most 180 results per search.


Search tips


What's new since Chapter 1

Full consistency audit (September 2026)

All five sources were audited together, against the real source code, not against changelogs or comments. The audit checked how each source launches, how results are capped, how settings are loaded, and how each one returns results to the forum.

Two bugs found and fixed

  1. PDF search, when nothing matched (fixed in PDFsearch v7.5.2) A search with no matches returned its result in the wrong shape. On the forum you didn't see this, because the message was the same either way. Run from the command line, it showed "ISE ERROR" instead of "No matches." Inside ISEmega it was silently logged as a PDF-source failure. Fixed.

  2. Orphan checker couldn't start (fixed in orphan_check v7.4.2) The maintenance tool that looks for orphaned attachments failed every time it started. It was still asking for a setting that had been renamed during an earlier settings clean-up. Fixed.

One false alarm

An orpha* search that seemed to return "No results found" was re-tested and could not be reproduced. Run correctly, it returns 18 matching posts across 14 topics, the same post the forum's stock search finds plus more. No fix was needed.


Documentation

The documentation set was rewritten to cover all five sources consistently:

Document For What it covers
In-app Help Everyone How to search, sort modes, and a section on how ISEmedia differs
Operator Manual Admins Install, troubleshooting, sorting internals, bug history
Python Scripts Manual Developers Code-level reference for every script, including the media indexer and the orphan checker

Reporting problems

If a search gives you something unexpected, please reply in this thread with:

Screenshots help a lot. The false alarm above was cleared up quickly because of one.


DMS/NX — Chapter 2. Continues the ISE Master Post (Chapter 1).