Skip to content
PYD ENTERPRISES

Source control · Ingest and export

PYD ENTERPRISES and Bitbucket

Ingest legacy repositories from Bitbucket Data Center or Bitbucket Cloud with read-only credentials, and return specification bundles as pull requests for review.

Bitbucket logo

Bitbucket

Source control

Ingest and export
scan.yaml (Bitbucket source)YAML
1sources:
2 - kind: git
3 url: ssh://git@bitbucket.internal:7999/claims/claims-web.git
4 ref: release/2019.4
5 auth: { secret: bb-access-key-claims }
6 dialect: struts-1.3
7 - kind: git
8 url: ssh://git@bitbucket.internal:7999/claims/claims-ejb.git
9 ref: release/2019.4
10 auth: { secret: bb-access-key-claims }
11 dialect: ejb-2.1

Where Bitbucket fits in a legacy estate

Bitbucket Data Center is common in insurers and banks that standardized on Atlassian tooling years ago, and it often holds the middle tier of a legacy system: Struts 1.x web modules, EJB 2.x session beans, Spring XML wiring and the WSDL contracts between them. Bitbucket projects tend to mirror application ownership, which gives PYD ENTERPRISES a clean unit of scan: one Bitbucket project, one legacy system.

PYD ENTERPRISES clones each repository over SSH with an access key or over HTTPS with an HTTP access token that has repository read permission. The mirror lives on a PYD ENTERPRISES volume inside your network and every scan is pinned to the commit it resolved at start. If a repository references another through a Maven parent POM or a shared JAR built in-house, list both repositories in the scan so the Java front end can resolve classes across them instead of reporting them as external.

Reviewing extracted specifications in pull requests

Exports go to a separate specification repository rather than the legacy code. PYD ENTERPRISES pushes a branch per scan and opens a pull request whose description summarizes what changed: operations added to the OpenAPI document, DMN decisions whose rules changed, JSON Schema properties whose constraints tightened, and rules newly flagged for human review.

Default reviewers and merge checks on the specification repository then work as they do for code. A common setup puts business analysts on dmn/ and architects on openapi/ and schemas/, so neither group has to read the other's artifacts to approve a baseline.

Data Center versus Cloud

With Bitbucket Data Center inside your perimeter, clone and push traffic never leaves your network. With Bitbucket Cloud, the scan workers need an egress rule to the Bitbucket hosts for clone and, optionally, push. Code is cloned into your environment and analyzed there; PYD ENTERPRISES sends nothing else out. Teams with egress fully denied can supply a git bundle exported from an approved host instead.

FAQ

Bitbucket questions.

Does PYD ENTERPRISES install an app or plugin into Bitbucket?

No. It uses ordinary Git access with credentials you issue. Nothing is installed on the Bitbucket server.

Can one scan span several Bitbucket repositories?

Yes, and it should when they form one application. Listing them together lets PYD ENTERPRISES resolve classes, Spring beans and WSDL bindings across repository boundaries.

What if the legacy code depends on JARs we no longer have source for?

PYD ENTERPRISES records them as binary dependencies with their package names. Calls into them show up in the dependency graph as external boundaries, and any rule that depends on their behavior is flagged for review rather than guessed.

See the Bitbucket flow on your own code.

Run PYD ENTERPRISES on 100,000 lines of COBOL or Java in your private VPC and review OpenAPI specs and DMN tables tomorrow.