80s right through 90s
I was taught Pascal at uni (Southampton) then found myself at my first job programming in Pascal for the next 15 or so years for the Quantel Paintbox family.
It was embedded, so we made our own hardware, so cross-platform wasn't an issue. We used the Oregon Pascal-2 compiler, running on DEC Vax and outputting for 68000 family. That compiler had the best of both worlds - Pascal levels of type checking, but a selection of features useful for real systems (eg separate compilation, the ability to turn off bits of checking as required, and the ability to use pointers to fixed addresses to allow us to write to our hardware's registers).
We wrote our own real-time OS for it. We had the spec for how to get Pascal to call assembler or vice versa.
Down-sides? Lack of IDE or debugger - development done with text editor (initially Dec EDT, later VaxTPU with our own extensions), debugging done by writes to console (or with a logic analyzer if you're dealing direct with hardware). Sometime mid-90s we started being able to edit using PCs and send to the Vax to batch compile.
At the time (early 80s) when this direction was chosen I think C was too new and didn't give you the help Pascal type-checking did. Don't let anyone tell you the compiler should do whatever the programmer asks because quite often the programmer is wrong on some details so a bit of help from the compiler is very helpful. And when you found you did need to defeat some checking you could do.
We had the source for one version of the compiler, Oregon having disappeared without trace, and made the occasional mod to it. Eventually the rest of the world caught up and we moved to C++, though there were attempts at other ways forward, such as trying to compile the Oregon Pascal compiler with Vax Pascal, or gpc, or with trying to combine the concepts of cross-gcc with gpc to get a cross-gpc.