Repository navigation
Description
Before You Submit
- I'm using the latest version of Salesforce CLI
- I've searched the existing issues
- I've run
sf doctorto diagnose common issues
Related Issue
No response
Command
sf project deploy start
Org Type
Sandbox
Did this work before?
Yes, this is a regression
Last Working Version
2.156.0
Summary
Since @salesforce/source-deploy-retrieve 13.8.3, every Profile is converted through the new ProfileMetadataFinalizer. When the root <Profile> element declares a namespace besides the default one (for example xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance", used by xsi:nil), that declaration is removed from the root and written as a child element named <@_xmlns:xsi>. The org then rejects the file with "Error parsing file". Any Profile carrying such a declaration now fails to deploy.
Repository to Reproduce
No response
Steps to reproduce
- In any Salesforce DX project, create
force-app/main/default/profiles/UsesNs.profile-meta.xml:<?xml version="1.0" encoding="UTF-8"?> <Profile xmlns="http://soap.sforce.com/2006/04/metadata" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"> <custom>true</custom> <description xsi:nil="true"/> <userLicense>Salesforce</userLicense> </Profile>
- Create
force-app/main/default/profiles/DeclaresNs.profile-meta.xml:<?xml version="1.0" encoding="UTF-8"?> <Profile xmlns="http://soap.sforce.com/2006/04/metadata" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"> <custom>true</custom> <description>Declares xsi but does not use it</description> <userLicense>Salesforce</userLicense> </Profile>
- Run
sf project convert source --metadata Profile --output-dir outand inspectout/profiles/*.profile. - Or run
sf project deploy start --dry-run --source-dir force-app -o <sandbox>.
Expected Result
The xmlns:xsi declaration stays on the root <Profile> element, as it did up to CLI 2.156.0 / SDR 13.4.6:
<Profile xmlns="http://soap.sforce.com/2006/04/metadata" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
<custom>true</custom>
<description xsi:nil="true"/>
<userLicense>Salesforce</userLicense>
</Profile>On 2.156.0, the dry-run deploy succeeds.
Actual Result
The declaration is dropped from the root and appended as a child element (the output of convert source is identical to what deploy start sends; you can confirm with SF_MDAPI_TEMP_DIR):
<Profile xmlns="http://soap.sforce.com/2006/04/metadata">
<custom>true</custom>
<description xsi:nil="true"></description>
<userLicense>Salesforce</userLicense>
<@_xmlns:xsi>http://www.w3.org/2001/XMLSchema-instance</@_xmlns:xsi>
</Profile>The deploy fails for both profiles. This happens over both SOAP and REST (SF_ORG_METADATA_REST_DEPLOY=true).
Error Message
DeclaresNs Error parsing file: The content of elements must consist of well-formed character data or markup.
UsesNs Error parsing file: The prefix "xsi" for attribute "xsi:nil" associated with an element type "description" is not bound.System Information
{
"architecture": "darwin-arm64",
"cliVersion": "@salesforce/cli/2.156.8",
"nodeVersion": "node-v24.14.1",
"osVersion": "Darwin 24.6.0",
"rootPath": "<npm global>/node_modules/@salesforce/cli",
"pluginVersions": [
"@oclif/plugin-autocomplete 4.0.5 (core)",
"@oclif/plugin-commands 5.0.5 (core)",
"@oclif/plugin-help 7.0.5 (core)",
"@oclif/plugin-not-found 4.0.5 (core)",
"@oclif/plugin-plugins 7.0.10 (core)",
"@oclif/plugin-search 2.0.5 (core)",
"@oclif/plugin-update 5.0.8 (core)",
"@oclif/plugin-version 3.0.6 (core)",
"@oclif/plugin-warn-if-update-available 4.0.4 (core)",
"@oclif/plugin-which 4.0.5 (core)",
"@salesforce/cli 2.156.8 (core)",
"agent 2.2.3 (core)",
"apex 4.4.4 (core)",
"api 2.0.19 (core)",
"auth 6.0.0 (core)",
"data 5.1.19 (core)",
"deploy-retrieve 4.2.18 (core)",
"info 4.0.18 (core)",
"limits 4.0.6 (core)",
"marketplace 2.0.18 (core)",
"org 7.0.2 (core)",
"packaging 3.0.14 (core)",
"schema 4.0.12 (core)",
"settings 3.0.17 (core)",
"sobject 2.0.13 (core)",
"telemetry 4.2.1 (core)",
"templates 57.4.8 (core)",
"trust 4.0.23 (core)",
"user 6.0.3 (core)"
],
"shell": "zsh"
}Additional Information
Resolved @salesforce/source-deploy-retrieve: 13.10.0. The same failure occurs on linux-x64 / node 24.22.2 with deploy-retrieve 4.2.24 installed as a user plugin (the hardisgroupcom/sfdx-hardis-ubuntu CI image).
Cause: mergeProfiles() in src/convert/convertContext/profileMetadataFinalizer.ts, added in forcedotcom/source-deploy-retrieve#1854:
const merged: JsonMap = { [XML_NS_KEY]: XML_NS_URL };
...
if (key === XML_NS_KEY || value === undefined) continue;Only @_xmlns is treated as a root attribute. Every other root attribute key (@_xmlns:xsi, @_xmlns:n1, …) goes through the merge loop like a child tag. It is collected into an array, and JsToXml serialises arrays as repeated child elements, so the result is <@_xmlns:xsi>…</@_xmlns:xsi>. Nested attributes such as xsi:nil survive because they sit inside a child object, which is why the prefix ends up unbound. The source-side merge path in finalize() (profile: { [XML_NS_KEY]: XML_NS_URL } plus the same XML_NS_KEY skip) looks to have the same problem. I didn't test that path.
Suggested fix: treat every key starting with the attribute prefix (@_) on the root Profile object as a root attribute: copy it to the merged output once, without merging it as a child.
Why "Last Working Version" is 2.156.0: from 2.156.1 onward, @salesforce/cli no longer publishes an npm-shrinkwrap.json. A fresh install of any later CLI therefore resolves the latest SDR (currently 13.10.0), even though deploy-retrieve 4.2.18 only requires ^13.4.6.
Workarounds:
- Install
@salesforce/cli@2.156.0, and remove any user-installed@salesforce/plugin-deploy-retrieveso that the core plugin is used. - Or strip the extra namespace declarations (and the
xsi:nilelements) from Profiles before deploying.
Screenshots or Log Files
No response
Activity
git2gus commentedon Nov 2, 2026
This issue has been linked to a new work item: W-24745383
WillieRuemmele commentedon Nov 2, 2026
Hi @tom-gangemi - thanks for reporting this, we'll get this fixed!
fix: keep Profile root namespace attributes when merging @W-24745383@ (…
WillieRuemmele commentedon Nov 2, 2026
Hi @tom-gangemi - we've published the fix, and with another bug we're working on closing, you should be able to just re-install the
latestCLI, and get this fix