TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
In a report on danluu.com, Dan Luu argues that software developers can now attempt projects that once seemed impractical, citing database work by Michael Malis and his own AI-assisted experiments. He says faster development does not remove the need for testing, benchmarking or product-quality work.
Software engineer and writer Dan Luu is arguing that developers should attempt more ambitious projects as new tools make some previously impractical work faster to build. In a report on danluu.com, he points to pgrust, a PostgreSQL-related project by Michael Malis, and describes his own small experiments as examples of what can now be attempted—while warning that speed does not establish software quality.
Luu says Malis has been showing him work on pgrust since earlier in the year and urging developers to be more ambitious. Luu reports that testing the project has exposed defects in PostgreSQL, including bugs involving numerical operations, JSONPath results, pgcrypto, memory safety in a dictionary loader and money overflow. These are Luu’s account of bugs found during testing; the source does not provide independent verification or a complete record of their status.
The argument is not that pgrust is finished or bug-free. Luu explicitly says it may not succeed. He describes its main pitch as performance and says some demonstrated optimizations could potentially inform work on PostgreSQL or ClickHouse, although that would be less direct than defects already reported and fixed. He also recognizes why readers may dismiss software rewrites: low-quality projects and unsupported performance claims have become common.
Luu gives examples from his own work: a regex engine adjusted for particular workloads, a ripgrep fork using that engine, and a Rust compiler-related project intended to speed builds. He says AI tools reduced his personal effort on experiments to a low single-digit number of hours, but distinguishes that from making a dependable product or public project, which would require substantially more work.
More Projects May Be Within Reach
The report matters because it frames faster AI-assisted development not simply as a way to produce more code, but as a way to attempt work that developers might previously have ruled out. Luu points to possible gains in developer tools and systems software, areas where faster builds, searches or file operations can affect repeated work across a project.
His examples also show why feasibility and readiness are different questions. A prototype can be useful for one person or workload without being reliable, broadly applicable or safe for others. Luu says reaching a useful product or open-source quality bar would require more effort than producing his experiments. For readers evaluating new software, the practical takeaway is to look for testing and workload-specific benchmarks rather than treating a rapid build or a claimed speedup as proof of quality.
Products Worth Considering
Logitech MX Mechanical Wireless Illuminated Keyboard, Tactile Quiet, Graphite
Logitech MX Keys S Wireless Keyboard, Graphite
DJI Osmo Mobile 8 Phone Gimbal with Mic Mini Transmitter, Infinity Black
As an affiliate, we earn on qualifying purchases.
The Case Behind the Claim
Luu ties the argument to growing skepticism about projects produced with AI coding tools. He says many rewrites attract suspicion because software that does not work well—and benchmarks that do not support their claims—are commonplace. He says he initially would have made a similar assumption about pgrust without his familiarity with Malis or a closer look at the testing.
He also describes how faster code-generation models could shift the bottlenecks in his own work. He identifies searching with ripgrep, Rust compilation, and system-call and filesystem overhead as areas that could become more noticeable when model responses arrive much faster. He reports that moving development from Mac to Linux improved some of the overhead for him, and speculates that specialized file-handling approaches could offer further gains. Those observations are personal experience and proposals, not evidence that the same changes will benefit every developer or system.
Luu’s wording is deliberately framed as an argument for ambition, not a claim that every large rewrite is worthwhile. The source says he might phrase the idea as “we can be more ambitious,” while Malis had used “we should” in informal conversations; Luu cautions readers not to read too much into that distinction.
“We should be more ambitious with the software we build.”
— Michael Malis, as quoted by Dan Luu
Prototype Gains Need Validation
The source does not establish whether pgrust will become a successful or production-ready project, or whether its performance will hold across representative workloads. It also does not give independent benchmark results or a detailed account of the PostgreSQL defects’ current status. Luu’s reported findings are evidence of testing activity, but they do not show that pgrust itself is free of bugs.
The broader claims about AI-assisted development are also limited by the examples provided. Luu describes his own time spent and particular workloads, not a controlled comparison across developers or projects. The source does not specify a publication date, the precise models used for the experiments, or the work needed to maintain and support the resulting tools. Those details would be needed to judge how widely the experience applies.
Testing Will Set the Bar
The report does not announce a release date, formal evaluation or next milestone for pgrust. The immediate test for the project and similar efforts is whether they can show reproducible performance results, thorough testing and dependable behavior beyond the creator’s own setup. Any PostgreSQL fixes or adoption of optimization techniques would need to be confirmed through the relevant project work.
For developers considering larger experiments, Luu’s argument is to attempt work that has become more feasible, then judge it by evidence rather than novelty or speed of creation. Whether these projects become useful tools remains open; the source’s clearest confirmed point is that Luu sees lower development effort as a reason to try more, not as a substitute for engineering discipline.
Key Questions
What is the report’s main argument?
Dan Luu argues that advances in development tools make some ambitious software projects more feasible. He says developers should still distinguish quick experiments from reliable products.
What is pgrust?
It is a PostgreSQL-related project by Michael Malis. Luu describes its main pitch as performance and reports that testing it has uncovered bugs in PostgreSQL.
Does Luu say pgrust is ready for production?
No. Luu says the project is not necessarily bug-free and may not succeed. The source does not establish that it is production-ready.
How much time did Luu spend on his experiments?
Luu says the experiments took a low single-digit number of hours of his time with modern tools. He says reaching a useful product or open-source quality level would take considerably more work.
What remains unverified?
The source does not provide independent benchmarks, a full account of the reported PostgreSQL bugs’ status, or evidence that Luu’s experience generalizes to other developers and workloads.
Source: hn
