Comment 33 for bug 1878234

Fabiano,

> I'm sorry you feel an easy way for distros to start using your
> changes right away is discussing the sex of angels.

No, I was *complaining* that we were discussing the sex of angels.
In other words, instead of (a) in comment #27, we have so far only discussed (b) and (c) which are IMO secondary.

> It's not, having the options being set during build time would
> just ease the adoption of the fix by the distros.

My patch allows that, unless I am mistaken.

> I'd like to ask you, please, be humble enough to understand
> people who more experience in the project may take their time
> to review your comments, to make suggestions, and that the
> suggestions my be different than what you proposed (or not).

My humility or lack therefore is even more off-topic than whether sysadmins with root access can be trusted with regexps, and veers even more into "personal opinion" territory.

> If it happens to be the case, please, be humble enough to listen
> to the ideas, try to learn from them, instead of forcing yours.

The regexp idea that you don't like was in response to Peng Tao's comments. While I did not immediately implement his suggestion as is (more specifically, I tried, it became very complicated, so I looked for another way), I recognized that he had pointed out a real issue, namely that my initial proposal would disable a feature in use cases which I consider legitimate. In other words, I learned from Peng Tao, and that's why the second iteration of the patch uses regexps and not string matching.

> Anyways, it's been made clear my suggestions are not valid nor
> welcome here.

If this is how you feel, then it's a wild misinterpretation of my response to your comments. You don't like regexps, fine. I share your concerns regarding the risk, so if there was no reason for them, I would not have added them.

I've explained the reason in comment #25, right after you brought the issue up in #24. Let me repeat here the core objection I have to removing regexps:

Until someone can prove to me that there is no QE system somewhere dropping nightly builds of qemu in a known location and testing them through an annotation, I'm reluctant to not being able to offer some wildcarding mechanism.

In other words, it's not that I don't hear you or don't learn from you. It's that I see a problem with your proposal, which you have not addressed. If you respond to my question, i.e. if you show me how a string list can be used to address the use case above, or if you show that the use case is not valid, then you will change my mind. Writing to my manager will not.