[vlc-devel] code conventions : line width ?

[Pierre d'Herbemont]

On Fri, Sep 12, 2008 at 12:48 PM, Rémi Denis-Courmont
<rdenis at simphalempin.com> wrote:
>
> On Fri, 12 Sep 2008 11:03:33 +0200, Laurent Aimar <fenrir at via.ecp.fr>
> wrote:
>>> 79 characters not counting the newline character (yes, some places do
>>> not follow this).
>>  I tend to dislike it as nowdays screen are much wider. It prevents using
>> variable with explicit names if you want to comply with it (otherwise you
>> end up using only a few words per line ;).
>
> Long lines are not readable. Regardless of the width of your screen. Why do
> you think real book (and TeX) use such short lines?
>
> In any case, I do use the console quite a log, and it still is 80 character
> wide. Also, if you have both hands on the keyboard, scrolling a long line
> is painful.
>
>>> That being noted I really dislike some of the conventions (Our Hungarian
>>> notation is sometimes harmful. The parenthesizing is insane for not
>>> matching what about every else agrees on).
>>  I strongly disagree with you on both points.
>
> You can disagree with factual truth if you like (I don't).
>
> As regards the Hungarian notation, the use of psz for not only string
> pointers, but also character arrays is confusing (Microsoft uses psz and sz
> respectively). The use of (p)p_foobar and i_foobar in the same struct is a
> confusing. block_t is a total disaster with p_buffer and i_buffer instead
> of (p_)data and (i_)length - in buffer management, you move the data
> pointer and the useful length, not the buffer pointer and the buffer size!

+1

> But my main objection to the systematic Hungarian notation is the i_
> prefix. It is CLEAR from the VLC source code, that it is used for any type
> of integers, effectively negating the whole benefit of Hungarian notation
> (to avoid type mismatch errors). Worse, it has caused widespread use of
> "int" instead of any more appropriate type (especially unsigned and
> size_t). This has caused a bunch of bugs, some security related, and
> remains a rampant problem to this day in the code base.  i_ is also used in
> case where it is totally pointless, e.g. file descriptors, which are
> abstract descriptors before they are integers.

+1