The Case for Windows

Why Adding Windows Support to Harper Desktop Is Sensible

Despite all my com­plain­ing about Windows Update and how bloated their soft­ware tends to be, Microsoft has suc­cess­fully con­vinced most of the world to use Windows.

At the time of writ­ing, 72% of peo­ple with a desk­top com­puter have Windows on it.

“Erm, ac­tu­ally Elijah, that’s not how mar­ket share works.”

Okay, fine. If we look into the fine print of the sta­tis­tic, it seems that it is in fact 72% of web user agents who de­clare them­selves to be run­ning on Windows. Close enough.

That’s a ton. To put that in con­text: For every ma­cOS user (which ac­count for 15% of all users), there are nearly five Windows users. That’s a pretty huge gap De­spite how much I’d like to see Windows put in its place, it’s a su­per com­mon op­er­at­ing sys­tem (and a pretty good one for most peo­ple, if I am to­tally hon­est).

To put it in in­vestor-speak: The TAM is plainly way big­ger than ma­cOS.

When we first started our work on Harper Desktop dur­ing Radical Speed Month, we only built it to sup­port ma­cOS. Why only ma­cOS, given this bounty of in­for­ma­tion about OS mar­ket share? Hon­estly, it was be­cause only 50% of the team (one of two, to be pre­cise) had a Windows de­vice, and we both had ac­cess to ma­cOS de­vices. Given that it was only sup­posed to be a month-long ex­per­i­ment, we felt com­fort­able not stress­ing about it.

Today, though, the desk­top ap­pli­ca­tion is a big part of (what I imag­ine to be) Harper’s fu­ture. So to­day, I want to make the case that we need to add Windows sup­port to Harper Desktop. I write this post, not just for my­self, but also to shed more light on the road map for those who have been ask­ing me when we will be sup­port­ing Windows.

The Cost of Adding Windows Support Is Low

Beyond be­ing some­thing we should do, adding Windows sup­port is some­thing that will be easy to do.

When we were in the rapid-pro­to­typ­ing phase of Harper Desktop, we de­cided to use Tauri for vir­tu­ally all fron­tend code and winit + egui for all the re­main­ing ren­der­ing. Since all three of these pack­ages are plat­form-ag­nos­tic, they will likely need no more than a few tweaks to work on Windows. Simple stuff.

Any code that touches the ac­ces­si­bil­ity API would need to be rewrit­ten for Windows. Fortunately, all of that hides be­hind the OsBroker trait. We would only need to im­ple­ment a few (admittedly com­plex) func­tions.

Once the pro­ject it­self is done, I es­ti­mate the mar­ginal main­te­nance cost to be pretty low. Com­pared to ma­cOS, main­tain­ing Windows ap­pli­ca­tions is a breeze. Sys­tem APIs rarely change, and the de­vel­oper tool­ing is miles ahead. Unlike ma­cOS, Microsoft does­n’t dep­re­cate in­dus­try stan­dards.

In fact, I be­lieve that adding Windows sup­port will have neg­a­tive mar­ginal main­te­nance cost, since it will make it far eas­ier to de­bug plat­form-ag­nos­tic is­sues. Bugs we find and fix in Windows (due to Harper’s ar­chi­tec­ture) will also be fixed on ma­cOS. Pesky per­mis­sion and ren­der­ing prob­lems, be­gone!

The People Have Been Asking for It

Ever since we put out the ma­cOS app for Harper, I have been re­ceived a tor­rent of mes­sage ask­ing for Windows sup­port. The peo­ple want it. Let’s give it to them.