[Issue 10866] Regression (2.064 git-head) Massive compiler slowdown

[d-bugmail at puremagic.com]

http://d.puremagic.com/issues/show_bug.cgi?id=10866



--- Comment #19 from Dmitry Olshansky <dmitry.olsh at gmail.com> 2013-10-12 01:34:24 PDT ---
(In reply to comment #18)
> (In reply to comment #17)
> > I think our only way out is to isolate the unicode_tables is separate
> > compilation + ~ .di file so that compiler doesn't have to re-tokenize 1.3M
> > file.
> 
> Please do so!

I stand corrected as I just tested it.
This is not a fix, it would be just another huge regression: everything in
std.uni would stop being CTFE-able at all. This cascades to things such as
toLower, isGraphical and such that are being used everywhere. 

It would also destroy my effort to free std.regex from reinventing its own
unicode stuff internally.

I'm going to investigate what exactly takes that much time

e.g. 
dmd -c std\internal\unicode_tables.d

Takes around 110 ms for me, quite a lot but not a showstopper. That means both
tokenization and compilation was performed and yet even half a second is not
seen.

-- 
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------