Hi,

as I said at that time, any such hook would be a) quite expensive on
commit time, b) hard to implement correct. Think of a 'cws rebase', you
commit stuff here which someone else introduced, what if some files had
an override originally? Break on commit than? Separate commits with and
without override? Hardly acceptable. Commit rebases with override always
enabled? Then people will introduce CR/LF due to merge conflict
resolution - thinking that the tooling will prevent them from doing
this. Either the hook is able to classify every file correctly or we
can't really use it. Of course this can be done, but I would rather like
if Björn uses his time for a more sensible project other than beating a
dead horse :-)

We had the problem with the -kb flag for a long time in CVS and I hated
every minute of it (and with CVS at least there was a concept such as a
file is text or binary, SVN doesn't really know such a thing). Before I
go that way again I'll rather fix a thousand more *.cxx with CR/LF in
them. These files will at least not kill a release.

With hg I'll make the win32text extension mandatory for everyone who
edits files on Windows. Well, I do not trust the win32text heuristics
completely and I can't check if people really use the extension of
course, but I'll personally nail everyone to the next window cross
(virtually only) who still commits CR/LF in large amounts :-) Oh and
I'll run a local hook on the integration repository to find the
offenders early :-)

Thinking of that: maybe a local tool run at integration time might be
the solution for subversion as well. Together with a strict refusal to
integrate such a CWS even if it's one day before freeze date. Better
this than to make committing real binary files an unnecessary pain for
everyone.

Regards,
   Heiner

Jan Holesovsky wrote:
> Hi Heiner,
> 
> On Monday 31 of August 2009, Jens-Heiner Rechtien wrote:
> 
>>> it seems to me that lately a huge amount of lineend-changes did occur.
>>> Maybe it is just a bad impression because many of the cws I had a look
>>> at lately did cause so many unrelated changes, but still:
>>>
>>> *Please* take care of lineendings before you commit.
>>>
>>> especially: please don't create files with mixed lineendings.
>> I can only emphasize how important this is.
> 
> Bjoern kindly offered to do a Python version of a pre-commit hook once; it 
> was 
> more complex, but can we at least try the CRLF check?  Bjoern, would you be 
> willing to provide that, if Heiner agrees to use it?
> 
> [IIRC, the main concern was the use of the overriding mechanism, but I still 
> think it would be better than nothing, more so if the hook provides an exact 
> info what to do to fix the situation.]
> 
> Thank you,
> Kendy
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [email protected]
> For additional commands, e-mail: [email protected]
> 
> 

---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to