Inform Compiler 636 is on ifarchive.org

General
1

I mentioned this in an Inform thread, but it seems big enough to mention here, as well. It was a real shot in the arm for me and the sort of positive surprise that lasts for a nice while!

It may seem like just a number, but it’s a big one. If anyone has a good-sized project where things compile … except … you see “Increase USE_MAX_TABLE_PROP_SIZE” or whatever, 636 makes this error redundant! It seems faster, too.

I’m not affiliated with anyone who did the hard work here, but I wanted to post this so others knew and also to say, thanks a ton! It decreases the overhead for putting out maintenance releases I’ve meant to create for a while.

Release notes
Executable(s)

According to this page it looks like we have @DavidG @DavidK, @mulehollandaise, @KrocCamen, and @zarf (in alphabetical order) to thank for the code changes. If I forgot anyone, please add them!

7 Likes
2

Yeah, I found it to be somewhat faster. About 10% faster when compiling a large I7-generated game.

I don’t remember exactly what made the difference. I think I optimized some of the string-compiling logic.

2 Likes
3

With my 2 (probably) largest games, I saw a drop from 6 seconds compile time to 4 (interpreting from I7 to I6 was at 9 seconds) and from 21 to 16 (interpreting was at 22 seconds.) These aren’t terribly scientific, but I wanted to lay out numbers to give strong evidence that whatever optimizations you did had nontrivial impact.

Both games have several large tables of random strings in them, so it might make sense that they had relatively more speed-up since your work was with string-compiling logic.

1 Like
4

I’ll have the Inform6 for Unix package updated in a couple days. This seems like a good time to pull together the Standard Library for an update.

4 Likes