I'm happy to see that people are using and benefiting from PSPP. I don't personally have time to maintain additional code, so--entirely apart from the provenance through an LLM--I can't in good conscience accept a large contribution of code that doesn't also come with the promise of maintenance.
If you have PSPP bug reports or bug fixes, I would like to receive them. I'd be pleased to hear that this is going to be made available separately from PSPP, and link to it from the PSPP web pages as long as it's provided under an appropriate free software license (such as the GPL). There is at least one PSPP web interface project, called Statable. You might want to join forces with them. Best wishes, Ben. On Mon, Nov 9, 2026 at 8:04 AM Jiří Čepelka <[email protected]> wrote: > Hello PSPP developers, > > My name is Jiří Čepelka. First of all, thank you for PSPP. > > Why I built this > > It started at my university, where we wanted a user-friendly, free > alternative to SPSS for my co-students. PSPP has an excellent statistical > core, but for people who are new to statistics it was missing some of the > analyses they need, and its interface looked dated and was hard to find > your way around. I wanted to see how far a modern, approachable interface > on top of the PSPP engine could go. > > What came out of it: PSPP Studio > > It is the PSPP 2.1.3 engine compiled to WebAssembly, with a new interface > that runs entirely in the browser (nothing to install, and no data ever > leaves the user's computer). It is available in Czech and English, and > PSPP's own messages are translated into Czech too. > > - Live demo: https://pspp-studio.vercel.app/ > - Source code (GPL-3.0-or-later): https://github.com/jirkacepelka/PSPP > > My question > > I do not have the time to maintain this on my own. Would the PSPP project, > or GNU more broadly, be interested in adopting and maintaining it? It does > not have to stay a web application: the engine and the design are > separable, and the interface could just as well be rewritten as a native > desktop application (for example to replace the current PSPPIRE GTK > interface), or reused in some other form. I care more about the idea of a > friendly free SPSS alternative living on than about the technology it is > built with. > > What it adds on top of PSPP > > - dialogs for the common procedures that generate and show the PSPP > syntax, next to a syntax editor > - a data grid with variable view, value labels and a variable inspector > - extra analyses that PSPP lacks, implemented in the browser on top of the > engine (for example GLM with repeated measures and profile plots, line > charts, and others) > - plain-language explanations of results, with effect sizes, aimed at > beginners > - import and export of CSV, Excel, OpenDocument and JSON next to > .sav/.zsav/.por, and export of the output to PDF, Word and HTML > - a guided first-run setup and a modern, accessible design (light and dark > themes) > > About the engine > > My changes to the upstream sources are small, and are in engine/pspp (see > "git log -- engine/pspp"): an optional --without-cairo build, an > html_table_to_string() helper, and Emscripten portability fixes. A small C > bridge (engine/bridge) exposes the library to JavaScript, and the build > scripts are in engine/scripts. The compiled engine is committed so the app > can be built without Emscripten. > > What I would like to know > > 1. Would you be interested in taking this in, as part of PSPP (as a web > interface, or as a model for a new native interface) or as a separate GNU > package? If not, I would be glad of any suggestion for where it could go, > or which of its parts would be worth contributing separately (for example > the Cairo-optional build, or the extra analyses together with their unit > tests). > 2. If yes, what would be needed from my side: copyright assignment to the > FSF, following the GNU coding standards (the front end is TypeScript/React, > not C), or something else? I am happy to do the handover work. > > Being open about how it was made > > Most of the front-end code, and the bridge and build scripts around the > engine, were written with the help of an AI model (Anthropic's Claude Opus > 5.5), under my direction and review. I understand that some free software > projects have reservations about AI-assisted code, for licensing, > provenance or other reasons, and I fully respect that. If this is a concern > for you, I understand completely, and I will keep the project separate and > not submit any of that code upstream unless you explicitly ask for it. I > would rather you know it before deciding than find out later. > > Thank you for your time. > > Best regards, > Jiří Čepelka - University of Hradec Králové FF >
