Comments on: Grep cache

By: Chris J

Chris J — Fri, 06 Nov 2009 13:41:55 +0000

I’ve been trying a few things out and resorted to trying this method (rejected before due to size of codebase, but lets see how it goes).

Just as a heads up, you might be able to do away with some cygwin’y stuff and speed things up with native Windows tools (removing the overhead of cygwin – which IME can be slow when dealing with lots of stuff). Instead of ‘grep’ look at ‘findstr’; similarly for ufind, try ‘dir /s /b’. I’m still using Perl though (Strawberry Perl)

Thus I’m creating my index with:
findstr /n /s /i /e . *.asp *.cs *.cls *.xml *.xsl *. …. | perl -pe “s/[ \t]+/ /g” > ..\index.idx … I’ll then use native-NTFS to compress the file and findstr to search it

By: Ignas

Ignas — Fri, 22 Jun 2007 14:49:59 +0000

1) Ah, right, I forgot you’d moved to Emacs. Does Emacs not do a binary search on the tags file (like, um, Vim does)? Or does it do something really silly like load it all into a buffer and then search it?

IIRC no it does not do a binary search, and yes it suerly loads it and searches it.

I am using GNU id-utils for a fast search. It is a nice indexed text search utility, not sure if there is a windows version though.

By: Larry Clapp

Larry Clapp — Mon, 18 Jun 2007 21:47:23 +0000

2) Lots of “Render”? That’s stupid. You should get them to fix their code. (Yes I’m kidding.

Anyway, best of luck.

By: bradbeveridge

bradbeveridge — Mon, 18 Jun 2007 13:53:54 +0000

I haven’t tried the ctags multiple file approach. I tried a single ctags file on our 14Mb code base. I think I had two problems
1) Searching was very slow – this is probably a function of Emacs’ tags lookup more than anything else
2) Our C++ code has many classes that have the same member function names – names like Render. I seem to remember ctags not dealing very well with that
I’ll probably try ctags again when my frustration levels get high enough.

By: Larry Clapp

Larry Clapp — Mon, 18 Jun 2007 12:39:20 +0000

Brad,

Thanks for posting this. I finally got around to reading it.

It sounds like an interesting tool. I’d probably go with sean’s suggestion and put the filename and line number right in the “database”.

That said — have you read the ctags FAQ? (http://ctags.sourceforge.net/faq.html#15) It has several tips for using ctags in a directory hierarchy. Combine that with Vim’s ability to use multiple tag files, or even search an entire directory hierarchy for tags files, and I think ctags would work for you again.

If that doesn’t work, I bet the ctags maintainer would like to hear from you.

By: krishna

krishna — Mon, 11 Jun 2007 16:20:41 +0000

Tried OpenGrok? OpenSolaris uses it for cross referencing. It works well with large code bases.

By: phil

phil — Mon, 11 Jun 2007 03:22:30 +0000

Have you tried tagbrowse?r It’s a gui tool (Win, Mac, Linux) for searching withing exuberant ctag files. It can search tag files of 10’s of MB in less than a second.

http://www3.telus.net/ngan/tagbrowser/tagbrowser.html

If you like vim and use visual studio you should also check out viemu.com. The nice thing is that you still get the facilities of visual studio as well as vim editing functions.

– Phillip
(author of tagbrowser)

By: mike

mike — Sat, 09 Jun 2007 10:24:22 +0000

you have seen the “Find References” entry on the “VAssistX” menu, haven’t you?

By: bradbeveridge

bradbeveridge — Fri, 08 Jun 2007 21:36:48 +0000

The Perl code used to be broken. I had a <= sign – WordPress ate some of the code after that. There may still be breakages.

By: sean

sean — Fri, 08 Jun 2007 20:52:16 +0000

Another approach might be to prepend the filename and line number when creating the index, effectively faking the “grep -n” output that Emacs uses:

find src … | xargs awk ‘{ print FILENAME “:” FNR “:” $0 }’ > index