svn commit: r210479 - head/usr.bin/grep

[Doug Barton]

On Sun, 25 Jul 2010, Gabor Kovesdan wrote:

> Em 2010.07.25. 22:17, Doug Barton escreveu:
>> On Sun, 25 Jul 2010, Gabor Kovesdan wrote:
>> 
>>> Author: gabor
>>> Date: Sun Jul 25 18:57:48 2010
>>> New Revision: 210479
>>> URL: http://svn.freebsd.org/changeset/base/210479
>>> 
>>> Log:
>>>  - Fix -l and -L by really surpressing output and just showing filenames
>>>
>>>  Submitted by:    swell.k at gmail.com
>> 
>> .... and dougb. :)  (It was part of the patch I sent on Friday.)
> I had to merge your patch with another one that I had in my working copy and 
> I didn't catch this part. Sorry for that, it was my mistake.

Everyone makes mistakes, sure. My argument here is that A) grep is too 
important a utility for mistakes like this to be acceptable, and B) your 
mentor should have caught this.

>> And not to pile on, but regression testing should have caught this. First 
>> by you, and if all else fails, by your mentor. At very least the thought 
>> process could have gone something like: "Let's see, the patch that I/Gabor 
>> generated to fix this problem is different than the one Doug generated. Now 
>> Doug is not a _complete_ moron, so let's take a careful look at this change 
>> and see if it really does what we think it should do."
>
> Nobody told that you were a moron

YOU didn't, nor did your mentor, however your statement is not literally 
correct. :)  Seriously though, that was just a bit of humor.

> and I did run regression testing, the test suite that GNU grep has.

I think it's great that you did that, however given that you were 
changing the behavior of the -q and -l options you should ALSO have set 
up a simple test file and done:

grep blah testfile
grep -l blah testfile
grep -q blah testfile ; echo $?
grep -ql blah testfile ; echo $?
grep -qL blah testfile ; echo $?

and made sure that everything worked as expected ... and so should your 
mentor.

> Regression testing is good but sometimes real users are necessary to 
> axe out bugs

I already both conceded, and offered you my sympathy on this point. :)

> and I think we are going into the right direction and I still don't 
> consider importing BSD grep a mistake.

I never said that importing it was a mistake. I said making it the 
default before it had wider testing was a mistake, but that we have yet 
to see whether it was a mistake that needs to be corrected or not.

And just to be clear, the reason I'm speaking up about this is that 
there is a risk that the NEXT project that comes along and gets put into 
HEAD as the default right away actually DOES break a whole lot of stuff. 
It may very well be that making bsdgrep the default from the start was 
acceptable. But nothing would have been seriously harmed if the typical 
procedure were followed, and the bugs that were found would only have 
impacted those who chose to opt in.

> Bugs might appear but as you can see I'm trying to fix them asap

Yes, I think you've done a good job of being accessible and responsive, 
thanks for that.

> and grep is improving in this way. Although grep seems to be a little 
> silly utility on the top of the regex library, it's quite complex and 
> it involves a lot of things,

As I've said before, I actually agree with you on this point. It's also 
one of those things that you don't realize how important it is until 
it's not working quite right. :)  You've got nerves of steel for taking 
this project on in the first place.

> Let's not flame about it any more just concentrate on bugfixing.

Just to be clear here, I'm not flaming at all. I'm not calling you 
names, or insulting your parentage. I'm expressing what I think are 
legitimate concerns that apply not only to this project, but to the 
paradigm of introducing similar projects down the road. And to be even 
more clear, I think that introducing more bsd-licensed utilities is a 
great thing, and I'm very excited to see that there are more such 
projects in the works. That makes it even MORE important to make sure 
that we all understand the importance of introducing them into the base 
in a manner that is minimally disruptive.


Doug

-- 

 	Improve the effectiveness of your Internet presence with
 	a domain name makeover!    http://SupersetSolutions.com/

 	Computers are useless. They can only give you answers.
 			-- Pablo Picasso