Repository navigation
pyegeria 6.2.32: ops-command fixes, Parent ID keeps memberships, templates as package data, dashboard resolver - #581
Merged
Merged
Conversation
…_specs pyegeria shipped three copies of the generated report specs as Python code: pyegeria/, pyegeria/config/ and pyegeria/view/. All were last generated before 6.2.28, and nothing in pyegeria imports them. The two orphan copies, and the pyegeria/config/ package that held one of them, are removed. pyegeria/view/dr_egeria_reports.py is regenerated: 304 sets, the same as md_processing/data/generated_format_sets.json; the old file had 150. refresh_specs now regenerates it on every run (step 3c, gen_report_specs --emit code), and that step is idempotent. gen_report_specs --emit code writes there by default. The view README says what the file is. Test: the file's set names equal generated_format_sets.json's, the orphan copies are gone, and refresh_specs runs the step. The test-order dependence noted in 6.2.28 (test_rs_egeria_tier's allowlist test after test_format_set_executor_resolution) no longer reproduces at 6.2.31 in any of three orders, so no change was needed for it. Release note: docs/release-notes/6.2.32.md. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
…t they say (ISSUE-175..179)
From Mandy's review of the operations commands (6.2.30 and a live quickstart),
each re-checked on 6.2.31.
- ISSUE-175: `hey_egeria_ops show integrations --targets` was registered as
"--targets", which click reads as an option. It is now "targets", with
function integrations_targets, which no longer shadows integrations_status.
In egeria.py the command had the same shadowing name and an argument
declared as "--connector"; both are fixed.
- ISSUE-176: `tell engine-host refresh` declared a single-dash
-engine-host-guid option and ignored it.
- --engine-host-guid is now passed as server_guid. Without it, the
configured engine host is used by name.
- --engine passes gov_engine_name.
- refresh_gov_eng_config gains organization_name and resolves a name as a
resource name. The old "Engine Host::<name>" qualifiedName lookup finds
nothing live.
- ISSUE-177: `tell servers start` looked the server up by property "name",
which finds nothing live, and passed the result as display_name. It now
calls activate_server_with_stored_config(display_name=...,
organization_name=...). --server is required.
- ISSUE-178: `tell servers stop` resolved qualifiedName == "<server>", which
never matches. shutdown_server (sync and async) gains display_name and
organization_name and resolves by resource name; the command uses it.
--server is required, and the commented-out options are removed.
- The three RuntimeManager methods resolve through _resolved_server_guid. It
raises PyegeriaInvalidParameterException when no server matches, instead of
sending "No elements found" as the GUID. Checked live, read-only:
resourceName "Coco Pharmaceuticals.qs-engine-host" and
"....qs-metadata-store" resolve; "name", "displayName" and the old
qualifiedName lookups do not.
- ISSUE-179: add_catalog_target / update_catalog_target (sync and async)
defaulted permitted_sync=BOTH_DIRECTIONS and delete_method=ARCHIVE. Every
add therefore overrode the connector's registered setting, and every update
reset both. Both now default to None and are left out of the request.
Tests (argument level, fake clients, nothing reaches Egeria):
- the arguments each command passes;
- --server is required;
- the name is resolved as a resource name;
- an unknown server is refused before any request;
- catalog-target fields are absent unless given;
- the targets command is reachable.
The ISSUE-156 test now expects the fields to be absent.
Release note: docs/release-notes/6.2.32.md.
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
- commands/ops/README.md is rewritten for the modules that exist: the
monitor_* / list_* displays, refresh_integration_daemon,
restart_integration_daemon, load_archive and gov_server_actions, with their
console scripts and hey_egeria_ops commands. It also lists the two legacy
modules (ISSUE-180).
- PYEGERIA_ISSUES.md:
- ISSUE-175 to ISSUE-180 are added;
- the entry-point note now says `hey_egeria_ops tell servers start|stop`;
- two Egeria-side follow-ups from the same review are recorded, with no fix
here: PostgresConfigurationProperty.REPOSITORY_MODE is typed DATE, and two
handlers read the metadata collection QN where the metadata source QN is
meant.
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
The basic/advanced command templates rendered by gen_md_cmd_templates now
ship in the wheel as md_processing/data/templates/{basic,advanced}/<Family>/
<Command>.md, so a consumer (Egeria Advisor first) reads templates that match
the installed Dr.Egeria executor instead of a cloned copy that drifts.
- One source of truth: sample-data/templates stays the generator's output and
the only copy in git. setup.py's build_py hook copies basic/ and advanced/
into the build tree; MANIFEST.in puts them in the sdist (uv build makes the
wheel from it). The hook fails the build if they are missing, so a wheel
never ships without templates.
- md_processing.templates: list_templates(level, family=None),
list_families(level), find_template / read_template(command, level,
family=None), templates_root(level), templates_source(level) ->
"package" | "checkout" | "missing". Reads importlib.resources first,
falls back to <repo>/sample-data/templates in a checkout or editable
install. Command lookup is by heading or file name, case/underscore
tolerant.
- Verified: wheel holds 436 basic + 458 advanced templates; in a clean 3.13
venv the accessor reports source "package" with the same counts; editable
install reports "checkout". micro-tests pass.
Brief: BRIEF-EA-CATCHUP-AND-DASHBOARDS-2026-11-10 B2 (D2).
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
…d membership _sync_parent_relationship reads the element's relationships of the parent type at either end and deletes the last one that isn't the named parent. On a folder that is one of its own members' CollectionMembership links. Severity HIGH (silent membership loss). Repro, cause and proposed fix (add only, child end only, page to the end) in the entry. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
…SSUE-181) _sync_parent_relationship read the element's relationships of the parent type at either end and deleted the last one that wasn't the named parent, so a Parent ID on a folder removed one of its own members' links. It now: - reads only links with the element at the child end (startingAtEnd 2/1, plus each entry's elementAtEnd1), paging to an empty page; - does nothing if the parent is already linked; - otherwise links the parent, leaving any other parent link in place and naming it in the output. Moving an element is an explicit remove. Micro-tests cover the folder repro (server filtering ends or not), the no-op, paging, a kept different parent, and a parent at end 2. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
…B4 / D4) pyegeria.view.dashboard_resolver.resolve_dashboard_sheet(client, name, *, perspective=None, run=False) resolves an Egeria-held Dashboard Sheet as the viewer the client is signed in as, so the Portal and Egeria Advisor resolve a sheet the same way. Data only: no rendering, caching or auth; Egeria stays the only server. Each placement comes back in sheet order with its position, kind, span/emphasis/perspectives/ detail spec, the Report read as the viewer (spec, outputFormat, params, analyticParams), the ReportSpec's FormatSet and source tier, runnable/refusal from pyegeria's spec resolver, and with run=True the exec_report_spec result. Kinds: report, text, missing, restricted. Restricted (§4.7): Egeria drops an element in a zone the viewer can't see from relationship reads and searches without a word, but an explicit read by GUID is refused with OPEN-METADATA-SECURITY-403-020 (OpenMetadataAccessSecurityConnector). So the writer now records the Report's GUID on each report placement field (additionalProperties reportGUID), and the resolver reads it by GUID: refused -> restricted, "not visible to you"; not found -> missing. An EGERIA-tier spec (possibly synced by another user) is re-read by its GUID as the viewer too. A refused token is re-raised, never read as restricted. A placement written before reportGUID was recorded, whose link is gone, can only say "missing". Also re-exported as pyegeria.view.dashboards.resolve_dashboard_sheet; listing per viewer via list_dashboard_sheets_for_viewer. Tests on a viewer-aware fake: visible, restricted Report, restricted ReportSpec, missing (gone and no recorded GUID), text, order and perspective, refused token, per-viewer listing, run=True, and the five live sheets resolving to the reader's placements. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
An app that caches its Egeria sheet reads (the Portal) hands the sheet it read to resolve_sheet / async_resolve_sheet instead of resolve_dashboard_sheet walking the chain again. resolve_dashboard_sheet now reads the sheet and calls it. Test: the sync form on an already-read sheet with a restricted Report gives the same kinds. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
… (ISSUE-181) If the page bound is reached before an empty page, report "could not read every <type> link; not added" instead of adding, which could duplicate the parent link. Architect read-back ask on 7e4fb20. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
Ops commands do what they say (ISSUE-175..180); Parent ID never deletes a relationship it does not manage (ISSUE-181). Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
Rendered Dr.Egeria templates ship as package data (build_py hook). Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
resolve_dashboard_sheet / resolve_sheet, the shared dashboard data step (B4). Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
One copy of dr_egeria_reports.py, regenerated by refresh_specs; removes the pyegeria/config package and the stale report-spec copies. Conflict: docs/release-notes/6.2.32.md was added on both sides. It is combined into one note covering odpi#580, the templates package data, resolve_dashboard_sheet, ISSUE-175..181 and this cleanup. Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
Signed-off-by: Dan Wolfson <dan.wolfson@pdr-associates.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
pyegeria 6.2.32: Mandy's operations-command fixes (ISSUE-175..180), Dr.Egeria Parent ID no longer deletes collection memberships (ISSUE-181), Dr.Egeria templates shipped as package data, a shared dashboard-sheet resolver, and removal of stale generated report-spec copies. Builds on #580 (My Profile as a production package).
Each part was read back by the Architect; the release branch merges the exact read-back heads (02c23ea, d05fb21, 640eb30, 84b0911) onto main be88d03, with the 6.2.32 release-note conflict resolved and a version-only bump.
Checks: full 3.13 micro-suite 2774 passed, 0 failed. Wheel in a clean venv: templates 436 basic / 458 advanced (source
package), my_egeria + serve_my_profile installed,pyegeria.view.dashboard_resolverimports,pyegeria.configremoved.Behaviour changes: catalog targets no longer default to BOTH_DIRECTIONS / ARCHIVE;
hey_egeria_ops tell servers start|stoprequire--server; Dr.Egeria Parent ID no longer moves an element (remove the old link explicitly);pyegeria.configno longer exists (usepyegeria.core.config).Release note
pyegeria 6.2.32
pyegeria 6.2.32 makes these changes:
pyegeria/configpackage are removed.New
move from
my_egeria/my_egeria/DemoCode/My_Profile/into themy_egeriapackage. The app module ismy_egeria.my_profile, somy_profileandserve_my_profilerun it. Diagnostic logging is off. TheUI is tidied: you can add a community, communities show persons rather than roles, screen titles are
consistent, and the edit-profile and Tech Types navigation are fixed. The docs and tests follow the
move. If you imported
my_egeria.DemoCode.My_Profile..., import frommy_egeriainstead.under
md_processing/data/templates/{basic,advanced}/<Family>/<Command>.md: 436 basic and 458advanced. A consumer (Egeria Advisor first) can read the templates that match the installed executor.
md_processing.templatesofferslist_templates,list_families,find_template,read_template,templates_rootandtemplates_source.templates_sourcereturns"package"when the templates come from the wheel,"checkout"in a checkout or an editable install, and"missing"otherwise.sample-data/templatesstays the only copy in git. Abuild_pyhook insetup.pycopiesthe templates into the build, and
MANIFEST.inputs them in the sdist. The build fails if they aremissing.
pyegeria.view.dashboard_resolver.resolve_dashboard_sheet(client, name, *, perspective=None, run=False)resolves a sheet as the signed-in user sees it, so the Portal and Egeria Advisor resolve itthe same way. It returns data only: no rendering, caching or auth.
restricted) and the Report and ReportSpec read as the viewer. With
run=True, the report's resultis included too.
rather than vanishing. The writer now records each report placement's Report GUID
(
additionalProperties.reportGUID) so the resolver can read it by GUID. A placement written beforethis release whose link is gone can only say "missing".
resolve_sheet/async_resolve_sheetresolve a sheet the caller hasSigned-off-by: Dan Wolfson dan.wolfson@pdr-associates.com
🤖 Generated with Claude Code