Do large numbers make Vim run slowly?

Question

This seems like a really daft question, but I have a Python file that has a number in it that's a thousand digits long and that file seems to be running very slowly, I'm not sure if there's some kind of processing going on that I'm unaware of.

I press line up j and there's a definite pause of ~1 second thats very painful!

I only have this on the Python file though, it doesn't happen on others.

After deleting the number the file works normally again...

Here is the number :

number=("""7316717653133062491922511967442657474235534919493496983520312774506326239578318016984801869478851843858615607891129494954595017379583319528532088055111254069874715852386305071569329096329522744304355766896648950445244523161731856403098711121722383113622298934233803081353362766142828064444866452387493035890729629049156044077239071381051585930796086670172427121883998797908792274921901699720888093776657273330010533678812202354218097512545405947522435258490771167055601360483958644670632441572215539753697817977846174064955149290862569321978468622482839722413756570560574902614079729686524145351004748216637048440319989000889524345065854122758866688116427171479924442928230863465674813919123162824586178664583591245665294765456828489128831426076900422421902267105562632111110937054421750694165896040807198403850962455444362981230987879927244284909188845801561660979191338754992005240636899125607176060588611646710940507754100225698315520005593572972571636269561882670428252483600823257530420752963450""")

It's not even a number really, its a string.

Answer 1

I can reproduce this with the syntax/python.vim that ships with Vim 7.4.663.

Using :syntime, this seems to be caused by the following syntax group / pattern:

TOTAL      COUNT  MATCH   SLOWEST     AVERAGE   NAME               PATTERN
73.870736   20     0       3.940215    3.693537  pythonNumber       \%(^\|\W\)\@<=\d*\.\d\+\%([eE][+-]\=\d\+\)\=[jJ]\=\>

You should report this problem to its maintainer (his name and email address is in the script's header).

Incidentally, I originally could not reproduce this, because I use an alternative syntax script from here. Switching to that (if it fits your requirements), would be a viable workaround, too.

---

Edit: Looking further into the probably cause, this seems to be due to pathological performance of the new NFA-based regexp engine. With :set regexpengine=1, I don't see that huge slowdown. So one possible workaround / fix would be switching to the old engine for that syntax match; this can be done by prepending \%#=1 to the pattern (cp. :help NFA).