I did my best not to introduce any kind of religious dogma into my argument. My problem isn't with the linux platform. My problem is in wondering how Jon's time could possibly be used like this and not impact heavily on his current market. And selfishly, on my favourite text editor.
Jon's in a sweet spot in Windows; ST is the prettiest editor on the platform. It's also extremely extensible. It appeals to a whole class of developers. I think it's ready for the big time, and he only needs to crack one market to do very well. Of course, there's stuff missing; people are evaulating ST right now and holding off buying it for that reason. If jon develops more features, more users cross over that threshold and choose to buy. But as I said at the start of my first post; this isn't a feature. If jon works on it, ST's feature set stays static. While it's static, those people out there evaluating ST are going to find that the product has halted, and not buy. Equally, Jon gets no closer to his V2.0 and to new licence income from existing customers.
A question is, how long would a port like this take? My intuition says that it's a great deal of work. Almost the entire core of ST would need to be rewritten. AFAIK, ST is a pure Windows beast; built on MFC and DirectX and targetted at file systems like NTFS. My guess is, it'd take about a year to port everything to one platform.
So, if I'm right, ST would get no new features for about a year. That's a year of reduced income, and a year in which competitors like 'e' and textmate make advances while ST stays still. In my book, that's a bad idea.
I said this because I've had my fair share of problems with scripts and packages that could easily have worked on windows, but contained little assumptions of linux. Just little things, like perl scripts assuming I had a HOME directory. Things that make the script fail.
Assumptions work both ways, of course; assuming that '\' is the directory separator, or that there will be a registry, or that symbolic links exist, or that your 'alt-x' keybinding will work when you don't have an 'alt' key.
However, it happens, it happens all the time, and it means that there will be addins that just don't work on your chosen platform.
I agree. How will a linux version not hinder Jon's effort to get these things in place?
It's enough that almost no-one does it. If it were cheap, TextMate would be available on Windows, and Photoshop on Linux.
Those products that do tend to build, not on Windows or Linux, but on another platfrom like the JVM or Mozilla's XUL. ST isn't a java product, so porting is just going to be hard. Porting MFC to Cocoa, or MFC to Gnome, is expensive and hard.
There's a nice article in Joel Spolsky's 'Best Software Writing' book, on the cost of porting Word 6.0 from Windows to Mac. It gives a bit on insight into the cost of moving pure windows app to pure mac apps. Have a read. It's almost directly applicable to this situation, and tellingly, it's an article that starts "Mom always said, 'the best thing about beating your head against the wall is that it feels good when you stop'"