svn commit: r256343 - in head/usr.sbin/bsdinstall: . scripts

[Teske, Devin]

On Oct 13, 2013, at 1:14 AM, Nathan Whitehorn wrote:

> 
> 
> On Sat, 12 Oct 2013, Teske, Devin wrote:
> 
>> 
>> On Oct 12, 2013, at 8:03 AM, Dag-Erling Sm?rgrav wrote:
>> 
>>> "Teske, Devin" <Devin.Teske at fisglobal.com> writes:
>>>> The code to replace the use of camcontrol is a a *very* complex parsing
>>>> of the geom XML configuration data stashed in sysctl. jmg@ started the
>>>> ball rolling on that.
>>> 
>>> You realize there is a text version as well?
>>> 
>>>> Yes. Which has been discussed at-length, you didn't need to put a
>>>> sandbag on my back (publicly no less; thanks for that).
>>> 
>>> Umm, I think Nathan was pretty civil.  You're the one who's turning this
>>> into a catfight.
>>> 
>> 
>> Reflecting upon the thread to see if you're _right_...
>> 
>> 1. He stated there were still some issues. [definitely civil]
>> 2. "I am surprised you committed it especially to stable/10,
>> before those issues were resolved." [civil? or inflammatory?]
>> 3. "I'm also not sure if people can review their own patches." [misunderstanding]
>> 4. "Installer regressions are very easy to introduce and very problematic
>> when created." [statements like that invariably lead people to believe he views
>> the commit as a regression -- I explained in a follow-up that it is not a regression]
>> 5. "Real review for installer changes is thus especially important this late in the
>> release cycle." [I read this invariably as he views that the commit did not go
>> through "Real review", but again... there is no regression and it's purely value-
>> add]
>> 6. "Do you have any plans to fix these issues in the very near future?" [definitely civil]
>> 
>> What got me ralled up was #'s 2, 4, and 5.
> 
> Hi Devin --
> 
> I'm sorry you felt I was attacking you.

It is what it is.


> I was, as I said, very glad to see someone work on ZFS support in the installer. The patch seems to have only minor issues, most of which were identified earlier. You will note that I am asking you to fix them, but not for a backout, and for you (and
> everyone else) to ask for review for future non-trivial installer patches,
> just as you would for changes to any discrete unit of the operating system.

Asking is the first step. You flat out attacked, imho.

I looked out over my castle walls and saw/see you at the gate in full armor. I'm not
going to let someone like that inside my courtyard.



> With respect to the patch itself, there are a few other architectural things
> that need to be fixed: the man page needs updating,

Yes indeed, it does.



> you did not update the
> part of the installer that does unattended installations, etc.
> 

Ought to be addressed for 10.0-BETA2 ... indeed I cannot disagree.

So that I don't forget, I'll put them here:

Need to document:

	ZFSBOOT_POOL_NAME=zroot
	ZFSBOOT_BEROOT_NAME=bootenv
	ZFSBOOT_BOOTFS_NAME=default
	ZFSBOOT_VDEV_TYPE=stripe
	ZFSBOOT_GNOP_4K_FORCE_ALIGN=1
	ZFSBOOT_GELI_ENCRYPTION=
	ZFSBOOT_GELI_POOL_NAME=bootpool
	ZFSBOOT_GELI_BOOT_SIZE=2g
	ZFSBOOT_GELI_KEY_FILE=/boot/encryption.key
	ZFSBOOT_DISKS=
	ZFSBOOT_PARTITION_SCHEME=GPT
	ZFSBOOT_SWAP_SIZE=2g

	Notice how all the variables specific to the `zfsboot' script start with ZFSBOOT_

As well as these two (they are not necessarily specific to the zfsboot script):

	nonInteractive=1
	zfsInteractive=1




> The end comment was not to say that this patch *is* a regression -- it clearly isn't since it adds a new feature and doesn't touch existing ones -- but a general comment that the installer is sensitive to them.

I've been working on the FreeBSD installer since 2005.
How long have you been working on it?

(full disclosure: that's not a back-handed comment... I really want to know...
I get the *distinct* impression that you think I'm some junior-level apprentice
on the matter considering the amount of times you have felt the need to tell
me "the installer is sensitive" -- you may have had your commit bit longer, but
that hardly means that you've put more work into the problem; the elitist
attitude needs to go)


> Unlike most parts of the system, it is run only rarely (developers updating their systems from SVN don't reinstall them from media) and is simultaneously the first thing and, until recently, the *only* thing, new users see when they first download FreeBSD.

You think I don't know this?



> So bugs can easily not be found but, when they are found, totally cripple all other features of the operating system since it can't be installed for anything else to even be used.

Please don't make me take the gloves off.



> Similar things can be true for things that are not quite regressions (new features that don't work as advertised, typos in menus) since the chance we notice them is very low while the chance new users do is high.
> 

Who anointed you king of the installer world?

In my mind, the corporate enterprise is *still* putting bounties out on the bugs you introduced
by regressing us from sysinstall to bsdinstall. I for one can't believe you introduced bsdinstall
in the state that it was in and further can't believe that it's still in the state that it is in for 10.0.
It's egregious, but perhaps worse is that you're standing in the way of progress.


> This is why the review requirements, and testing times, for installer changes are typically quite high. This is especially true in the run-up to a release.

Again with rhetorhic... you imply it wasn't reviewed nor tested.

I think you're angry because you weren't part of the process.
-- 
Devin

_____________
The information contained in this message is proprietary and/or confidential. If you are not the intended recipient, please: (i) delete the message and all copies; (ii) do not disclose, distribute or use the message in any manner; and (iii) notify the sender immediately. In addition, please be aware that any message addressed to our domain is subject to archiving and review by persons other than the intended recipient. Thank you.