When the analyzer is configured and running, it will return analysis results. These results will be transformed into data useful for the standard vscode flows and for any custom views.
Story Highlights
Two presentations of the results are possible:
Display incidents as standard file level lint errors (file diagnostics)
Display the set of issues with the affected files grouped under the issue
Other language servers and linters will only add problems to the list when a file is opened. This can make it difficult to find all of the problems without resorting to running a linting tool outside of vscode and linking all the errors back. With a project level analyzer running, the full incident->problem list can be populated right at the beginning. As files are opened, changed, and saved, the analyzer can update the results and the problems can be adjusted to the current state of the source file.
Standard file level lint error style just requires knowledge of the file, line, and column coordinates
For a single incident...
it shows up in the Problems view
severity (info, warning, error) is defined from the incident's issue's category
In a file editor...
at a glance, only incident will be visible
a single incident shows up as code problem/hint/underline
when code problem is hovered, at least the incident's message is shown
when code problem view is open, at least the incident's message is show
provide a link inside the code problem view to open the incident details
provide a link in the code action menu (Ctrl+.) to open the incident or issue details
potential to use a code lense / code lense sample to highlight where an incident is located with the basic info and action buttons presented (corollary is a git extension that can highlight the beginning of a change in the last commit)
provide some extension level configuration for:
the underline display styling
the scroll bar indicator block
code lense information
Status indicator for the workspace...
Count of issues (perhaps even by issue category) and incidents should be shown independently from the problem count status indicator
Hover over the counts show the same details but with words
Click on the counts opens the issues tree view, or panel, or full editor view
Status indicator for an open editor....
Count of issues and incidents for the open editor can be shown
Hover over the counts show the same details but with words
Click on the counts opens a quick pick menu that would have Kai actions relevant to the entire file. Could be something like:
Goto first incident
Reanalyze file
Ask Kai to fix all incidents in this file
A Konveyor AI View Container can setup the primary sidebar with analyzer and issue focused views
Tree view where data is listed by: Issue > File > Incident
Each issue is a root item on the tree
Each issue item can be expanded to view the files containing the incidents
Each file item expands to all of the incidents in the file for that issue
Issue view...
Could be a webview in the Editor area
Could be a table like view in the Panel area
Provide an issue list that is filterable and sortable
Like the basic tree view, opening an issue can show the linked files and incident
Unlike the basic tree view, opening an issue can show the description text and any other detailed information about the issue, including but not limited to incident count, files affected count, category/severity, effort score all above the nested table listing the file/incident details
Note: This can be similar content to the Issues view of the Konveyor web UI
Allow toggling details on and off (off would show details in a hover/popover)
Clicking on a file in the issue view will open the file in the editor and navigate to the first known incident in the file
Clicking on an incident will open the file in the editor and navigate to the selected incident in the file
Actions should be allowed to:
Ask Kai to fix an issue across all linked incidents
Allow developer ranking of issue (a custom sort order)
With the analysis forward approach, we will want to emphasize the issues over the incidents
Considerations
Analysis produces a set of issues. The results can be streaming or a single document with results for the entire project.
Two branches for interaction with incidents:
as file based problems in the normal linter/codemod pattern
as issues at the project level potentially spanning multiple files
Each issue contains data including:
ruleset containing the rule triggered
rule id of the rule triggered
name
description
category (mandatory, potential, ...)
effort (as an integer)
labels of the rule triggered
links set of URLs to reference for the issue
incidents one for each incident - if file contains multiple incidents for the rule, each incident will be listed separately with the file name repeated
Each incident contains data including:
file the is affected
line of the incident in the file
message markdown formatted message to describe the incident
codeSnip code snippet of the incident extracted by the analyzer
Surface analysis results
When the analyzer is configured and running, it will return analysis results. These results will be transformed into data useful for the standard vscode flows and for any custom views.
Story Highlights
Two presentations of the results are possible:
Other language servers and linters will only add problems to the list when a file is opened. This can make it difficult to find all of the problems without resorting to running a linting tool outside of vscode and linking all the errors back. With a project level analyzer running, the full incident->problem list can be populated right at the beginning. As files are opened, changed, and saved, the analyzer can update the results and the problems can be adjusted to the current state of the source file.
Standard file level lint error style just requires knowledge of the file, line, and column coordinates
For a single incident...
In a file editor...
Status indicator for the workspace...
Status indicator for an open editor....
A Konveyor AI View Container can setup the primary sidebar with analyzer and issue focused views
Tree View for issues...
Issue view...
Issuesview of the Konveyor web UIWith the analysis forward approach, we will want to emphasize the issues over the incidents
Considerations
Analysis produces a set of issues. The results can be streaming or a single document with results for the entire project.
Two branches for interaction with incidents:
Each issue contains data including:
Each incident contains data including: