> -----Original Message----- > From: lm-sensors-bounces at lm-sensors.org [mailto:lm-sensors-bounces at lm-sensors.org] On Behalf Of > Hans-J?rgen Koch > Sent: 18 July 2007 20:27 > To: lm-sensors at lm-sensors.org > Cc: Mark M. Hoffman > Subject: Re: No longer reviewing random hwmon patches > > Am Mittwoch 18 Juli 2007 20:14 schrieb Jean Delvare: > > Hi all, > > > > Twice in the past 2 weeks, I have been criticized for being too > > demanding in my hwmon patch reviews. I have been told that I was > > wasting developers' time and taking the fun away from them. Well, I am > > also no longer having fun reviewing patches when I am "rewarded" this > > way. Not that reviewing patches is really fun to start with, you > > usually don't receive much thank nor praise, but receiving criticism is > > going one step too far. > > I absolutely agree. And I don't think you're wasting anybodies time. > I remember when you sent me back my max6650 patch 16 times. I still > think that this was no waste of time. It is now clean code which follows > a clean concept. A waste of time would have been to merge my first > version. It would have wasted the time of an unknown number of unknown > programmers in the future, who would have to deal with bad code > and violations of the hwmon sysfs specifications. > > So, as I said before, I thank you for _your_ time and that you > made it possible for me to contribute a piece of _good_ code so > _my_ time was not wasted. > I absolutely agree. If I am to run code on my systems, I want to be sure that it has been thoroughly reviewed by whoever was responsible. When I submitted my vt8231 driver, Jean was very thorough with his review, and I *appreciated* this. One of the problems with Linux's wider acceptance is that more and more people want to get involved. This isn't a bad thing in itself, but just because someone wants to help doesn't mean that their contribution must automatically be accepted. The reason Linux is growing in popularity is because it is so reliable. People who want to have their code should recognise this and accept that the standards for acceptace of code are high and are set by the reviewer, not the submitter. If you cannot take criticism of your code when you want to get it submitted into a larger system then it is you that has the issues, not the reviewer or the review process. I don't want to run code on my systems that has been through some half-assed or incomplete review process. I want to be confident that the code is written well and reviewed properly. With Jean's attitude and dilligence, the review process was done properly. He never criticised code for the sake of it - every criticism was valid and relevant. > > > > It has been suggested that I should review patches differently and > > accept more divergence from what I consider the right way. This isn't > > going to happen, sorry. Given that I am not paid for the job, nobody > > gets to tell me which patches I should review and how. But I also don't > > want to cause contributors to leave the project. > > If a "contributor" wants to leave because his crappy code is rejected, > I say: Good riddance! hwmon is a subsystem where most of the people > don't have the hardware to test a driver, so bugs can stay there for > a long time. That's why I find it especially important to have strict > reviews before something gets merged. Agreed. If you are a prima donna with your code and either too arrogant or stupid to accept valid criticism then I for one don't want to run it - I don't trust code written by someone with that attitude! > > > > > As a consequence, I have decided to change my behavior with regards to > > the patches which are posted to the lm-sensors list. I will no longer > > review random patches. > > I understand your point of view. > > > I will only review patches meeting any of the > > following conditions: > > * I asked for that patch. > > * The patch is for a hwmon driver I am maintaining. > > * The author explicitly asked me to review his/her work, and I feel > > like it. > > * Mark asked me to review the patch. > > > > Hopefully this will prevent further tensions from arising. > > > > Actually, I think this is a big loss for hwmon. I'd still be glad if > you reviewed my patches (there will be more soon). > > Thanks, > Hans > Jean, you have my support and my respect for your hard work and skills. If others around you don?t realise what you bring then it just shows how much they still have to learn. I just hope that lm-sensors' code quality doesn't slowly degrade over time if the review process is relaxed. Please Mark H., follow Jean's prior practice at enforcing thorough reviews before accepting code into the kernel. - Roger Lucas
