100 percent vs 85 percent.
Measured, not claimed.
On a 565-test Oracle benchmark — covering DBMS packages, DDL clauses, PL/SQL constructs, query syntax, and SQL functions — Trekport converts 100 percent cleanly. Ora2Pg converts 85 percent. The 15-point gap shows up most on DBMS packages (+26) and DDL clauses (+32), the areas where a Perl regex pipeline runs out of road.
Empirical benchmark — 565 Oracle test cases
We ran both tools through the same 565-test Oracle conversion suite (436 generic Oracle tests covering Oracle 12c–19c features, plus 129 aggressive stress tests covering Banner-derived patterns and edge cases). Each test asserts on token presence in the converted output and checks for residual Oracle syntax. Results below.
| Category | Trekport | Ora2Pg v25.0 | Δ |
|---|---|---|---|
| DBMS / UTL packages (61 tests) | 100.0% | 73.8% | +26.2 |
| DDL clauses (119 tests) | 100.0% | 68.1% | +31.9 |
| PL/SQL constructs (94 tests) | 100.0% | 92.6% | +7.4 |
| Query syntax (92 tests) | 100.0% | 91.3% | +8.7 |
| SQL functions (199 tests) | 100.0% | 92.0% | +8.0 |
| Total (565 tests) | 100.0% | 85.0% | +15.0 |
Methodology: same Oracle SQL fed to each tool; outputs scored against identical assertions; both tools running their default modes. Test corpus and runners live in the public repository at apps/engine/tests/oracle_coverage/. Trial-license verification welcome — request the full benchmark report from sales.
On a real Banner production sample: 8.7-to-1 win ratio
Synthetic benchmarks are useful for coverage. Production Oracle code is the real test. We took a 96-package Banner sample (the largest ERP Oracle estate by volume), converted with each tool, and deployed the output to vanilla PostgreSQL 18.3. The number is the percentage of packages that deployed with zero errors.
| Outcome | Count |
|---|---|
| Both tools deploy cleanly | 22 / 96 |
| Trekport wins (deploys, Ora2Pg fails) | 52 / 96 |
| Ora2Pg wins (deploys, Trekport fails) | 6 / 96 |
| Both tools fail | 16 / 96 |
| Trekport overall deploy rate | 77 percent |
| Ora2Pg overall deploy rate | 29 percent |
Where Trekport wins, it is usually because Ora2Pg's converted PACKAGE BODY emits as a single multi-procedure CREATE FUNCTION (which PostgreSQL cannot parse), or because typed REF CURSOR returns come out as CREATE TYPE x AS REFCURSOR RETURN y (not valid PG). Where Ora2Pg appears to win, inspection shows it ships stubs that compile but do not implement the original Oracle logic. Full per-package audit available on request.
Where Ora2Pg falls short
Ora2Pg is a respected open-source tool. It is also two decades old, implemented in Perl, and built around regex-based pattern substitution. Three structural limitations matter to anyone migrating an enterprise Oracle estate.
Regex-based conversion
Pattern matching breaks on the long tail. Compound triggers, package-level state, FORALL, BULK COLLECT, and circular package references all need manual review. The Ora2Pg docs say so explicitly. Trekport parses to an AST and emits structurally guaranteed PL/pgSQL.
Alphabetical export order
Ora2Pg exports tables alphabetically, not in foreign-key dependency order. The standard workaround is to disable foreign-key checks during load and re-enable after. Trekport runs a 14-phase dependency-ordered deployment that handles FKs natively.
Manual cleanup expected
The Ora2Pg documentation lists GRANT, TRIGGER, FUNCTION, PROCEDURE, TYPE, QUERY, and PACKAGE among the export types that 'have to be reviewed' after conversion. Trekport ships these deploy-ready as part of its 100 percent deployment mandate.
Capability comparison
| Capability | Trekport | Ora2Pg |
|---|---|---|
| Conversion approach | AST-based | Regex / templates |
| Measured 565-test pass rate | 100.0% | 85.0% |
| Banner production deploy rate | 77% (96-pkg sample) | 29% (same sample) |
| PL/SQL packages | Yes — deploy-ready | Manual review (per docs) |
| Compound triggers | Yes | Partial |
| BULK COLLECT / FORALL | Yes — set-based | Manual |
| Dependency-ordered deployment | 14-phase pipeline | Alphabetical, FK manual |
| Idempotent re-deployment | Yes | No |
| Validation reports | HTML + PDF | Cost-estimate report only |
| GUI | Native desktop | CLI (third-party GUIs) |
| Parallel data transfer on Windows | Yes | Disabled |
| Install | Single binary | Perl + DBI + DBD::Oracle |
| License | Commercial ($49/mo) | GPL, free |
When Ora2Pg is still the right call
If you have a small schema (under 100 tables, no PL/SQL), and your team is already comfortable with Perl, Ora2Pg is a solid free choice. We recommend it ourselves for that profile. The economics shift toward Trekport at the scales where manual cleanup time dominates the migration budget: PL/SQL-heavy estates, dependency- ordered deployments, regulated environments, and any project where calendar time matters more than tool license cost.
FAQ
Why is Trekport better than Ora2Pg?
Three reasons. First, Ora2Pg uses regex-based pattern conversion that fails unpredictably on PL/SQL packages, compound triggers, and the long tail. Trekport uses AST-based conversion (parses the source into an abstract syntax tree, transforms it, emits valid PostgreSQL). Second, Ora2Pg's own documentation states that PL/SQL functions, procedures, packages, and triggers must be reviewed manually after conversion. Trekport ships them deploy-ready. Third, Ora2Pg exports tables alphabetically, not in foreign-key order, so loads commonly fail at FK constraints. Trekport's 14-phase deployment resolves dependencies upfront.
Do I still need orafce after using Trekport?
Yes, just like with Ora2Pg. Trekport's converted output uses orafce for Oracle compatibility functions (NVL, DECODE, SYSDATE, DBMS_OUTPUT, etc.). Unlike EnterpriseDB's Migration Portal, both Trekport and Ora2Pg target vanilla PostgreSQL with the open-source orafce extension, not a paid Postgres distribution.
What does Ora2Pg leave for manual cleanup?
Per Ora2Pg's own documentation, the GRANT, TABLESPACE, TRIGGER, FUNCTION, PROCEDURE, TYPE, QUERY, and PACKAGE export types all need manual SQL editing post-conversion. On a 30,000-object enterprise estate, this typically lands as several thousand objects requiring DBA review. Trekport's 100 percent deployment mandate eliminates this manual work.
Does Trekport have a GUI?
Yes. Trekport Studio is a desktop application (macOS, Windows, Linux) with a project view, conversion progress, deployment dashboard, and validation report viewer. Ora2Pg is a Perl command-line tool by default, with various community GUIs (Cybertec Migrator wraps it visually, but adds licensing). Trekport ships a polished GUI in the box.
Can I run Trekport on Windows like Ora2Pg?
Yes, with one caveat that favors Trekport. Trekport runs identically on macOS, Windows, and Linux because it is a native desktop binary. Ora2Pg runs on Windows but disables parallel processing on that platform, so large data transfers are dramatically slower than on Linux. Trekport's parallel data transfer works the same on all three operating systems.
Run them both. Decide on output.
The Trekport free trial covers up to 10 tables, end-to-end. Run it alongside Ora2Pg on the same Oracle source. Compare the deploy output line by line. Buy the one that ships fewer TODOs.