The problem isn't really producing some kind of plot. There are some nice libraries that could do that for us, as you say. PSPP already has some plotting capabilities, although they could bear a lot of improvement.
The problem is that I'd like to support as much SPSS compatibility as I can. SPSS produces graphical output in a custom format called VizML, so compatibility entails supporting that same format; otherwise, PSPP can't display the graphs in SPSS output (this is currently a problem). And then, once PSPP supports that VizML format, which is actually a very adaptable and general-purpose description of graphical output, it will become natural to use it for PSPP itself. A lot of the work that I'm doing now is trying to understand the VizML format properly. It's rather complex. On Tue, Sep 1, 2026, 9:49 PM Friedrich Beckmann <[email protected]> wrote: > Maybe in that case the graphing could be easier by using things like > plotly? My > experience is that doing the plots „nice“ is quite some work, e.g. for > doing > the axis labeling. > > OpenROAD currently goes that way to move to a web based gui. But they often > run in containers where gui is a pain anyway. > > > Am 02.09.2026 um 06:37 schrieb Ben Pfaff <[email protected]>: > > > > I don't have a complete answer for that, but I have some thoughts. One > thought is that PSPP could use a web based graphical interface instead of a > gtk based GUI. In some ways, that's easier and more flexible these days, > and it doesn't have the same failure modes. > > > > On Tue, Sep 1, 2026, 9:26 PM Friedrich Beckmann < > [email protected]> wrote: > > This is really amazing to rewrite the whole pspp kernel in Rust! For me > the picture is > > not clear how the gui could be handled. Do you have any idea? Or do you > plan to keep > > the Rust version text only? > > > > > Am 01.09.2026 um 22:23 schrieb Ben Pfaff <[email protected]>: > > > > > > This year, I've spent most of the time that I can dedicate to PSPP > development on rewriting > > > PSPP in Rust. A big motivation for this is bugs: it is easier to write > reliable code in Rust than > > > in C, because of the way that the language is structured. Security > issues are closely related. > > > > > > But I've been able to do some work in the Rust version that I had not > done in the C version. > > > Currently, I am working on support for displaying graphs that appear > in SPV files. This will > > > ultimately allow PSPP to produce graphs in the full generality that > SPSS supports. I'm > > > making good progress on this. My hope is that I'll be able to release > this feature in the > > > Rust version of PSPP by the end of the year. > > > > > > I post about progress sometimes on my Mastodon account, which you can > follow at > > > https://framapiaf.org/@blp. I recently posted one example of progress > on PSPP graphs: > > > https://framapiaf.org/@blp/117197565164328276. You can also view > other PSPP > > > related posts that I've made there: > https://framapiaf.org/deck/@blp/tagged/pspp > > > >
