[TxMt] Re: Performance issues

[Paul McCann]

Aristide Grange wrote:

> Data: a 2.5 MB text file (plain text / no coloration / 225800  
> lines, none exceeding 30 characters) structured as follows:
>
>> ABACA~
>> ABACULE~
>> ABAISSABLE~
>> ABAISSAIENT~
>
> Problem: Replace all the 225800 occurrences of "~\n" by "\n" (iMac  
> G4 800 Mhz, 1GB ram, 10.4.6):
> 	TextWrangler: 30 s (gracefully interruptible)
> 	TextMate: who knows? (forced to quit after 15 mn)

I'm sure we've been through this whole thing before...
OK, let's make such a beast: if the stuff between the lines is  
"long.pl" then

./long.pl > long.txt

gives a suitable test file called long.txt
=======================================================================
#!/usr/bin/perl
foreach (1..60000){
print<<END;
ABACA~
ABACULE~
ABAISSABLE~
ABAISSAIENT~
END
}
=======================================================================
> Looks like there is some psychological issue here, indeed.
> ;-)
> I do love TextMate. But for such search/replacements, I couldn't  
> live without TextWrangler.

I can: I do. My withdrawal from BareBones is complete (I still own a  
current BBEdit license; well, I think it's current -- they might have  
tried to gouge another slab of upgrade cash from registered users  
while I haven't been watching). Anyhow... ignoring that(!).

Just be a tiny bit creative here: for example, if you want to search/ 
replace in a massive file such as long.txt use TextMate's integration  
with standard unix utilities: for example...

Text => Filter through command...

perl -pe 's#~\n#\n#'

On my eMac (similarly configured to your imac) it's about 12 seconds  
to do the replacement, and the command is no harder than any Find/ 
Replace dialog box. Yes, the built-in mechanism should probably  
behave better, but it's *really* easy to work around. Unless, of  
course, you're simply hunting for nits to pick.

Cheers,
Paul