For the design engineer

What you are actually worried about

Not features. You are worried that a restructure which takes ten minutes in a spreadsheet will take an afternoon of dialogs, that a half-finished assembly will be visible to a buyer before it is ready, and that at some point the tool will refuse to let you do something obvious and there will be no way around it.

Those are reasonable worries and two of the three have real answers. The third — being told no — is partly true by design, and the useful version of this page explains where and why rather than pretending it does not happen.

So: bulk editing, privacy of work in progress, and the specific set of things you will not be allowed to do.

Bulk editing that is genuinely bulk

Four hundred rows pasted from a spreadsheet produce four hundred correct lines. That is asserted by a test rather than hoped for, and it was written as a requirement before the grid existed.

Correct means typed. A quantity of 1,000 becomes the number one thousand, not the string — which would sort between 1 and 2 and break every rollup without warning. EA and ea normalise. Reference designator ranges expand and are counted against the quantity, and where the two disagree the row is raised as an exception rather than silently accepted.

Fill-down, multi-cell edit, bulk attribute change across a selection, and inline validation that explains a rejection instead of reverting the cell. Tree, indented and flat views over the same structure — three presentations of one answer rather than three queries that might disagree.

The layer that decides all of this has no DOM in it and is tested without a browser, which is why paste fidelity is provable rather than a thing that mostly works until a supplier sends an unusual export.

Work in progress stays yours

There are three levels on every object, and the one that matters to you is the third. Iterations are private to your workspace. You can save eleven times on a Tuesday, restructure an assembly across a week, and try something that turns out to be wrong, and nothing outside the workspace can reference any of it.

That is a stronger guarantee than hidden or draft-flagged. An iteration is not a referenceable object, so there is nothing for a report, a search, an export or an agent to accidentally surface.

Optimistic locking and check-out expiry handle two people and one part. If somebody else is editing, you find out before you start rather than when you save — and a check-out that was abandoned three weeks ago expires rather than blocking the team until an administrator intervenes.

Where the tool will tell you no

You cannot edit a released revision. Not with an override, not as an administrator. Something has referenced it — a BOM line, a drawing, a qualification, a unit that shipped — and editing it would silently change what those things mean. The route is a new revision through a change, which takes about a minute for a small change.

You cannot paste a row that does not reconcile without seeing it. A quantity of eight against three reference designators gets flagged. You can accept it if it is genuinely right — reference designators are not always complete — but you will have looked at it.

You cannot edit an ERP-mastered field. Cost, on-hand and lead time render greyed and labelled with the owning system, because the next sync would replace whatever you typed. That one frustrates people until the first time it saves them.

That is close to the whole list. Everything else — object types, attributes, lifecycles, workflows, numbering — is configuration your organisation controls and can change in a sandbox with a rollback available, rather than something a vendor has to do for you.

How it meets your ERP

Mostly it means you can see the consequences of a design decision without leaving the structure. Cost, on-hand quantity, open purchase orders and lead time read back from your ERP and appear against the lines, read-only, so a rollup tells you what a change costs before anybody has to ask procurement.

It also means the thing you release actually reaches manufacturing. On change release the resolved structure publishes outward, dated to the effectivity, carrying the packaging, firmware, labels and consumables that were never in the CAD model — which is the difference between a bill somebody has to fix and one they can plan against.

What it does not mean is more forms. Nothing about the ERP connection asks anything of you beyond raising the change you were going to raise anyway.

Where the boundary is

There is no SOLIDWORKS add-in. If you live in SOLIDWORKS, files check into the vault manually and metadata is entered or bulk-imported rather than read from the model. Onshape and Fusion connect over their APIs. That gap is real and it is the thing most likely to annoy you day to day.

There is also no in-app 3D viewing or geometry comparison. Thumbnails come from the CAD system's own API; comparing two revisions of a model is a question for the CAD tool, and Manufacturing PLM links out rather than rendering it worse.