On 6 Jun 2006, at 11:43, Mats Persson wrote:

Hi Milen,

I just saw your post and read it since it intrigued me the most out of the 1258 unread posts on this ML in my inbox ;)

On 5 Jun 2006, at 14:17, Milen Dzhumerov wrote:

...

Things which I miss:

• CodeSense: Basically this is my reason for not switching to

TextMate. There is an enormous amount of APIs on the Mac and I don't think developers should be expected to know each method and its arguments in their heads. Copying/Pasting from the Xcode documentation browser doesn't cut it - the time spent on this actually outweighs the benefits of using TM (for me personally). The day a CodeSense/IntelliSense/WhateverSense is implemented in TextMate you will have another customer. For now I'm putting my TM switch on hold.

I KNOW that this is NOT exactly what you are asking for or looking for, BUT this can be implemented today by yourself (& others). Allan is well aware of the CodeSense/CodeCompletion/etc requests and has been for a long time, so I'm sure he's implementing something like it back in his little toolshed ;-)

I assume quite a few people are requesting it.

When I was doing PHP development (never really done Cocoa) I created this bundle [ http://anon:anon@macromates.com/svn/Bundles/ trunk/Bundles/PHPCodeCompletion.tmbundle/ ] which you can see how I used with screenshots of at this outdated page [ http:// www.imediatec.co.uk/tm/phpcc/ ]

It looks quite nice and useful.

I have subsequently moved on to Rails development, and have therefore begun working on a something similar in Ruby, which you can find here [ http://www.imediatec.co.uk/tm/bundles/RubyAndRails- Experimental.tmbundle.v1.zip ]

The PHP bundle does more, but is implemented in PHP which really was/is a mistake compared to the Ruby/YAML versions, but never mind it worked/(works?).

Grab those two bundles and see how I - a NON-computer programmer - hacked my way to an improved coding experience in TM, and I'm sure that you can code your way around the perceived limitations of TM today exactly to your individual liking.

I'm going to do that as soon as my exams finish! Thank you very much for taking the time to reply to my email.

I'm fairly certain Allan and others has something going with regards to Cocoa/Objective-C that helps them speed up their coding, so look through the existing bundles and ask around.

Kind regards,

Mats

Milen

---

"TextMate, coding with an incredible sense of joy and ease"

• www.macromates.com -

"RubyOnRails development done the Mac way"

• locomotive.raaum.org -

---

For new threads USE THIS: textmate@lists.macromates.com (threading gets destroyed and the universe will collapse if you don't) http://lists.macromates.com/mailman/listinfo/textmate

On 7/6/2006, at 14:11, David Watson wrote:

Is there a way to get this to show the symbols for all files in a given directory?

No -- for that, TM would need to parse them all, which it doesn’t currently do. I am not entirely sure it would be the best approach -- e.g. something like ctags has dedicated parsers to only grab symbols, and which are much much faster.

[...] I recall this being common in IDE land, something like "Go To Definition".

This does sound much like ctags functionality.

On 6/8/06 12:10 AM, "Gerd Knops" gerti-textmate@bitart.com wrote:

...

...

[...] I recall this being common in IDE land, something like "Go To Definition".

This does sound much like ctags functionality.

Thanks, I wasn't aware of ctags.

Indeed, while ctags seems like the solution, my case is actually a bit more complex than it appears. I have to use this hybrid language that's a mix of lisp and C. One of the things that turned me on to textmate, was that I could just tell textmate to treat these files as lisp, and it behaved accordingly, simply ignoring the elements that it didn't understand. At least that was my impression. This works wonderfully.

The problem with ctags is that they try to infer the file type based on the extension, and some other bits described in their man page. I didn't see any way of simply telling ctags that "these are lisp files, act accordingly". The end result is that their parser can't deal with the hybrid mix of lisp and C that textmate handles so well. This is true even when I try to trick their parser by renaming all my files to *.lisp.

See here:

http://gerd.knops.org/?p=7

Thanks, I wasn't aware of TMCodeBrowser.

I tried the TMCodeBrowser plugin, but its dependence on ctags, and my dependence on the hybrid language described above, meant that the tag generation doesn't work for my language and so I couldn't experiment with TMCodeBrowser. It does look like exactly what I need, but I couldn't find any way to hack my sources such that the ctags parser would behave nicely, at least not without a lot of scripting work to remove the non-conforming elements, which may defeat the purpose.

More worrisome is the fact that textmate crashed after I installed TMCodeBrowser the first time. The crash occurred after I hit the relaunch button on the dialog. It seemed like the kernel was probably still alive (I didn't have another machine to ssh into it), but the desktop, keyboard and all GUI objects save for the mouse pointer were dead. That is, I could move the mouse pointer, but GUI objects were not responding to clicks, and the keyboard, including function keys and Cmd-Tab were dead. In addition, my menubar utilities showing CPU, memory, etc. had frozen. I waited several minutes before hitting the power button and the TMCodeBrowser install worked smoothly the second time, including the relaunch. I have the 1070 build of textmate on a 1.67 G4 powerbook with a gig of memory.

Regards, David