nimly Uses Too Many Loops In Compile-time

[loloicci]

No description provided.

[loloicci]

You can avoid this error to use the new compiler option maxLoopIterationsVM:N (maybe later than Nim v1.0.6) (See also https://nim-lang.org/docs/nimc.html).

Before v1.0.6, you can use niml with compiler build with vmdef patched as follows

diff --git a/compiler/vmdef.nim b/compiler/vmdef.nim
index 7a2bc1607..91706ae5d 100644
--- a/compiler/vmdef.nim
+++ b/compiler/vmdef.nim
@@ -17,7 +17,7 @@ const
   byteExcess* = 128 # we use excess-K for immediates
   wordExcess* = 32768
 
-  MaxLoopIterations* = 3_000_000 # max iterations of all loops
+  MaxLoopIterations* = high(int) #3_000_000 # max iterations of all loops

[loloicci]

improve some #25

[loloicci]

It maybe solved when nim becomes version 1.0.0.

[samdmarshall]

Hi, I'm encountering this issue using nim 1.0.6.

[loloicci]

@samdmarshall thanks for your reporting.
Could you share the code with the bug if you can?

[samdmarshall]

oh, sorry, yeah - of course. i'll try to isolate into a stand-alone sample. do you want me posting here in this issue or as a new one?

[loloicci]

Thanks!
Either is fine.
Link to #11 if you post as a new issue.

[bung87]

https://github.com/bung87/rust2nim I face this when compile entry file.

it's the first time I try to use a parser generator, I may doing something wrong, but this stop me get ing further information.

[bung87]

lexgen.nim line:378

if toSeq(sTran.keys) == toSeq(tran.keys): I modify to this , reduce the iterations, it seems under the compare calls the argument changes dynamically.
there's proc defaultAndOther called by lexgen.nim(567, 33) convertToLexData throw too many iterations error .

[loloicci]

@bung87 Thank you for your analysis!

if toSeq(sTran.keys) == toSeq(tran.keys): I modify to this , reduce the iterations, it seems under the compare calls the argument changes dynamically.

If you will create a new PR, I will merge this change after checking this has no unexpected effects.

You can avoid this error to use the new compiler option maxLoopIterationsVM:N (maybe later than Nim v1.0.6) (See also https://nim-lang.org/docs/nimc.html).

[bung87]

I will create a PR later, merge if you think that's correct since I has no much knowledge in this domain. write a wrong grammar also will cause this?

I checked after modified test will not pass.

[loloicci]

Sorry, @bung87.
Your suggestion does not work expectedly.

toSeq(sTran.keys) == toSeq(tran.keys)

This returns always true and causes error in nimble test.
(nim c -r tests/test_lexgen.nim provides more information)

I did some refactoring in lexgen.nim in the #40.
Check this if you are interested in what this part does.

[loloicci]

@bung87

write a wrong grammar also will cause this?

No (as far as I know).
This error is caused when the calculation needed to generate the lexer is too much.
Without bug in the implementation, this generation does not cause infinite loops.

[bung87]

Thanks for your explaintion! then I feel fine with maxLoopIterationsVM, from what I understand the lexer parser generate during compile time, that probably exceed the limit even code is fine, that sounds reasonable.

I just aware you removed the bug label, guess I can just try :P