Lots of seemingly unused defun's in csharp-mode

[josteink]

Amongst other the following private functions are defined, but does not seem to be used by any internal code:

• csharp--on-ctor-close-curly-p
• csharp--on-defun-close-curly-p
• csharp--on-namespace-close-curly-p
• csharp--on-enum-close-curly-p
• csharp--on-intf-close-curly-p
• csharp--on-class-close-curly-p

In this general area of the code almost all the defined functions are unused. Maybe this should be cleaned up?

[josteink]

I guess the initial job should be map out all unused functions to get the full picture about how many lines of code we're talking about here.

I guess taking out these functions, will again cause the list of actually used csharp-regexes in the regex-alist to be reduced as well.

I've git grepped across all revisions. These functions have never been used by csharp-mode code:

$ git grep "csharp--on-class-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-ctor-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-defun-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-namespace-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-enum-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-intf-close-curly-p" $(git rev-list --all) | grep -v defun
$ git grep "csharp--on-class-close-curly-p" $(git rev-list --all) | grep -v defun
$

So if they're not used, and we don't plan to use them, I really so no need to keep them around...

[josteink]

I did a big grep across all the code-base to see what things have never been referenced:

$ grep "^(defun " csharp-mode.el | perl -pe "s/\(defun ([a-z0-9-]+) .*$/\1/" | sort >defuns.txt
$ for line in `cat defuns.txt` ; do echo -n "$line: " ;git grep "$line" $(git rev-list --all) | grep -v defun | wc -l ; done >use-count.txt
$ grep ": 0" use-count.txt
c-inside-bracelist-p: 0
csharp-max-beginning-of-stmt: 0
csharp-move-back-to-beginning-of-defun: 0
csharp-move-back-to-beginning-of-namespace: 0
csharp-move-fwd-to-end-of-class: 0
csharp-move-fwd-to-end-of-defun: 0
csharp--on-class-close-curly-p: 0
csharp--on-ctor-close-curly-p: 0
csharp--on-defun-close-curly-p: 0
csharp--on-defun-open-curly-p: 0
csharp--on-enum-close-curly-p: 0
csharp--on-intf-close-curly-p: 0
csharp--on-namespace-close-curly-p: 0
$ 

That's going back through all revisions, so these are things which has never been used. Checking the current affairs, gives equal results:

$ for line in `cat defuns.txt` ; do echo -n "$line: " ;grep "$line" csharp-mode.el | grep -v defun | wc -l ; done >use-count.txt
$ grep ": 0" use-count.txt
c-inside-bracelist-p: 0
csharp-max-beginning-of-stmt: 0
csharp-move-back-to-beginning-of-defun: 0
csharp-move-back-to-beginning-of-namespace: 0
csharp-move-fwd-to-end-of-class: 0
csharp-move-fwd-to-end-of-defun: 0
csharp--on-class-close-curly-p: 0
csharp--on-ctor-close-curly-p: 0
csharp--on-defun-close-curly-p: 0
csharp--on-defun-open-curly-p: 0
csharp--on-enum-close-curly-p: 0
csharp--on-intf-close-curly-p: 0
csharp--on-namespace-close-curly-p: 0
$ 

Most of these seems privateish, and in like good candidates for stripping if we're not going to use them.

[wasamasa]

There's more:

• csharp-line-up-region/csharp-lineup-if-and-region (referencing each other only, not interactive)
• There's a few more movement commands that are not in use, but they all are interactive, so not worth removing them as users could be still binding them
• csharp-max-beginning-of-stmt
• I don't get what the c-forward-objc-directive advice is for. It seems like a thing better fixed in cc-mode itself if it is supposedly problematic. Appears to be used in c-guess-basic-syntax which was called "the easily most obfuscated code in Emacs" by quotemstr.
• Monkeypatching cc-mode generally looks like a bad idea to me, I'd like getting rid of that entirely. The problem is just that I don't know what exactly the monkeypatches addresses...

[josteink]

Not touching the monkey-patching as I too don't feel fully qualified to know if they're needed or not anymore.

Not touching interactive commands either.

Investigating csharp-line-up-region and csharp-lineup-if-and-region which supposedly controls indentation behaviour if the user decides to use them, I cannot see how anything in any way behaves differently, so I'm all OK for dropping those two.

With that done, this issue should be considered complete.

[josteink]

Closed by #74.