Dear coreboot folks,
Am Donnerstag, den 30.05.2013, 01:57 +0200 schrieb gerrit code review:
From http://review.coreboot.org/3336:
Change subject: gitconfig: Add sample checks shipped by git to pre-commit hook ......................................................................
Patch Set 2: Code-Review-2
It does not make sense to have a hook that checks a tree-wide policy and then allow that policy to be disabled by configuration. Either we decide on the filename policy or we do not need the hook.
as non-ASCII characters are not allowed in our source [1], I assume they are also not allowed in file names.
The sample pre-commit hook shipped by git has a check for non-ASCII file names, which can be disabled by a config option.
Are there any circumstances where non-ASCII characters might be needed? If not, Peter is right and the option can be removed from the script, which I would do then.
Thanks,
Paul
Paul Menzel wrote:
as non-ASCII characters are not allowed in our source [1]
Where the hell did you get the idea that there is such a policy?
You are extrapolating discussion about four characters in two files between two individuals to a tree-wide project-wide policy. That is absolutely unreasonable!
I assume they are also not allowed in file names.
Don't assume.. Especially not based on extrapolation like that.
The sample pre-commit hook shipped by git has a check for non-ASCII file names, which can be disabled by a config option.
Either we have a policy or we do not. If we have a policy then why do we care what the sample hook does?
Are there any circumstances where non-ASCII characters might be needed?
I don't think so and quite likely they would break things, so I would be in favor of a policy to only allow ASCII filenames.
If not, Peter is right and the option can be removed from the script, which I would do then.
You shouldn't have pushed a commit in the first place without strong consensus to back it up. Working together, communication, all that.
//Peter
Am Donnerstag, den 30.05.2013, 12:10 +0200 schrieb Peter Stuge:
Paul Menzel wrote:
as non-ASCII characters are not allowed in our source [1]
Where the hell did you get the idea that there is such a policy?
Because one of the project leaders reverted it and reading Ron’s commit message of the revert.
You are extrapolating discussion about four characters in two files between two individuals to a tree-wide project-wide policy. That is absolutely unreasonable!
Sorry.
I assume they are also not allowed in file names.
Don't assume.. Especially not based on extrapolation like that.
The sample pre-commit hook shipped by git has a check for non-ASCII file names, which can be disabled by a config option.
Either we have a policy or we do not. If we have a policy then why do we care what the sample hook does?
Are there any circumstances where non-ASCII characters might be needed?
I don't think so and quite likely they would break things, so I would be in favor of a policy to only allow ASCII filenames.
Alright. Good to know.
Does somebody object to such a policy?
If not, Peter is right and the option can be removed from the script, which I would do then.
You shouldn't have pushed a commit in the first place without strong consensus to back it up. Working together, communication, all that.
Sorry. I hope I fixed that by sending this message to the list. In pre-Gerrit times I would have tagged the subject with [RFC] (request for comments). And as patches can be discussed in Gerrit, I do not see why pushing it to Gerrit is a problem.
Thanks,
Paul
Paul Menzel wrote:
as non-ASCII characters are not allowed in our source [1]
Where the hell did you get the idea that there is such a policy?
Because one of the project leaders reverted it and reading Ron’s commit message of the revert.
That is still only one person. It's a good idea to take more input.
You shouldn't have pushed a commit in the first place without strong consensus to back it up. Working together, communication, all that.
Sorry. I hope I fixed that by sending this message to the list.
Yes, and thanks for taking it to the list. It's the right thing to do.
And as patches can be discussed in Gerrit, I do not see why pushing it to Gerrit is a problem.
Gerrit is a code review tool, not a communications system.
Email is a communications system, not a code review tool.
Code review requires careful attention to detail, communication perhaps slightly less so since there's a human on the other end.
Gerrit is a bad message board. Use the mailing list for discussion.
Thanks
//Peter