According to quality defect analyses conducted by the company using the Coverity tools, the overall inventory of 775,000 lines of code showed only 20 minor defects, which is equivalent to about 1 defect per 39,000 lines of code. By comparison, a Coverity analysis of the rival MySQL open source database conducted last winter showed 1 defect per 4000 lines of code.
For the record, both companies have or currently are fixing the defects, and contributing the patches back to the open source community. And both claim that the low defect rates are living proof of the ability of the open source model to tap QA pools dwarfing those of traditional commercial software vendors.
EnterpriseDB, which only announced its existence in May, currently offers a free beta for download today. Although it originally spoke of releasing the product this month, it decided to hold off until Linux World on August 9 to formally go GA with version 1, giving it time to add some additional features.
Those features included support of Solaris, Mac OS/X, and nearly all 64-bit and 32-bit Intel/AMD Linux distributions. It is also adding replication based on the open source Slony-I project, support of Oracle style stored procedures from Microsoft .NET clients, and user-defined functions.
According to company president Andy Astor, taking time to add features was in response to feedback from beta users. We changed plans because the product was stable enough, that the issues customers had were for new features.
The PostgreSQL engine comes from a code base that in some cases dates back 20 years. Although until now the system has had largely niche appeal, EnterpriseDB is striving to generalize the market with a commercially supported product that has Oracle compatibility.
Although the incumbent in the open source space is MySQL, Astor says EnterpriseDB’s real target will be Oracle, because of what he said is the proven scalability of the PostgreSQL engine.
