At the time, Knuth was exclusively using a PDP-10, "SAIL", running the WAITS time-share operating system. If you look at early ARPAnet maps, you'll see that maybe half of the machines were DEC PDP-20's; certainly at Stanford, MIT, CMU, etc., the first machines to be connected to (what was to become) the Internet were.
One main goal of the 1980's TeX rewrite was to make it as portable as possible, to support other popular architectures of the day. C wasn't available for DEC PDP mainframes; IBM 360s didn't have a C compiler at the time, either. PL/I, Algol, etc. were also not universally available. So, based on the fact that the Hedrick Pascal compiler existed for the PDPs, and a general notion that a reasonable Pascal was, or would soon, be available for VAX/VMS, IBM, and Unix (the gpc front-end for gcc), the choice was made. By the way, at the time it didn't seem like PCs and Macs would be viable platforms, being so memory-constrained (TeX wanted a whole megabyte to run in!) so that wasn't a big part of the consideration. And Unix workstations (Sun, HP, etc.) weren't yet really a thing yet.
It wasn't a super-comfortable decision, but seemed like the best choice at the time. Partly because of this, TeX is written to use just a subset of the Pascal language, so as to keep things simple, and thus ultimately translatable into other languages if necessary. It turned out that DEC eventually made a very good Pascal compiler for VMS, and similarly for IBM; but GPC lagged, and the popular Unix ports of TeX are based on C translations of TeX's Pascal code. Ditto for (all? some?) of the PC and Mac versions.
But, to your point, yes, all the various compilers of the time had their own work-arounds to the "default", "fopen", etc. problems. The fact that they chose different syntaxes wasn't a big issue, being easily addressable via the "WEB" macro language Knuth wrote on top of. Much more of a pain was Pascal's virtually useless string functionality and poor memory-management that required laborious work-arounds, resulting in code that is difficult to modify and debug. It's a real shame that standardization of the features needed to make Pascal a language that was really suitable for cross-platform production code didn't happen in time to make a difference. All of the code Knuth writes has been in C (via CWEB) for quite some time now.
[Edit: tried to say "star"-nix, but that made everything \it, so switched to "Unix".]
One main goal of the 1980's TeX rewrite was to make it as portable as possible, to support other popular architectures of the day. C wasn't available for DEC PDP mainframes; IBM 360s didn't have a C compiler at the time, either. PL/I, Algol, etc. were also not universally available. So, based on the fact that the Hedrick Pascal compiler existed for the PDPs, and a general notion that a reasonable Pascal was, or would soon, be available for VAX/VMS, IBM, and Unix (the gpc front-end for gcc), the choice was made. By the way, at the time it didn't seem like PCs and Macs would be viable platforms, being so memory-constrained (TeX wanted a whole megabyte to run in!) so that wasn't a big part of the consideration. And Unix workstations (Sun, HP, etc.) weren't yet really a thing yet.
It wasn't a super-comfortable decision, but seemed like the best choice at the time. Partly because of this, TeX is written to use just a subset of the Pascal language, so as to keep things simple, and thus ultimately translatable into other languages if necessary. It turned out that DEC eventually made a very good Pascal compiler for VMS, and similarly for IBM; but GPC lagged, and the popular Unix ports of TeX are based on C translations of TeX's Pascal code. Ditto for (all? some?) of the PC and Mac versions.
But, to your point, yes, all the various compilers of the time had their own work-arounds to the "default", "fopen", etc. problems. The fact that they chose different syntaxes wasn't a big issue, being easily addressable via the "WEB" macro language Knuth wrote on top of. Much more of a pain was Pascal's virtually useless string functionality and poor memory-management that required laborious work-arounds, resulting in code that is difficult to modify and debug. It's a real shame that standardization of the features needed to make Pascal a language that was really suitable for cross-platform production code didn't happen in time to make a difference. All of the code Knuth writes has been in C (via CWEB) for quite some time now.
[Edit: tried to say "star"-nix, but that made everything \it, so switched to "Unix".]