cvs commit: www/en/projects/ideas index.sgml

Sun Feb 18 14:55:06 UTC 2007

Quoting Robert Watson <rwatson at FreeBSD.org> (from Sun, 18 Feb 2007  
11:46:44 +0000 (GMT)):

>
> On Sun, 18 Feb 2007, Alexander Leidinger wrote:
>
>> Quoting Robert Watson <rwatson at FreeBSD.org> (from Sat, 17 Feb 2007   
>> 19:37:48 +0000 (GMT)):
>>
>>> On Sat, 17 Feb 2007, Alexander Leidinger wrote:
>>>
>>>>> - Magic symlinks: Several implementations exists, so we don't need more
>>>>> people looking at this right now.
>>>>
>>>> But we need people reviewing them and chosing the right one. So   
>>>> the  entry needs to be changed instead of removed.
>>>
>>> I think an alternative explanaation is that people have looked at   
>>> them and been left sufficiently worried by the experience as to   
>>> wonder whether "magic symlinks" are really a good idea.  I think   
>>> we should take it off the list before we get yet another set of   
>>> patches that won't be accepted for the same reason.
>>
>> There are mixed feelings about this in the responses. AFAIR it can   
>> be summarized to: If it is not enabled by default and needs to be   
>> activated even when compiled in (sysctl), then nobody will object.   
>> The crowd which is interested in the magic symlinks would be happy   
>> with this solution too.
>
> No, I disagree.  We will not accept security holes that are disabled by
> default if the primary purpose of the feature is to cater to
> environments in which the feature will be a security hole.  This is a
> property of at least one of the patches submitted to date, and my
> feedback along these lines was (I seem to recall) entirely ignored.
> Please stop asking people to implement magic symlinks unless you are
> willing to provide the necessary oversight to make sure that we don't
> get yet more patches that represent security holes.

Ok.

>> If an entry is removed completely because it is inappropriate we   
>> should list it somewhere and explain why it will not be accepted in  
>>  the tree.
>
> I think we shouldn't try to enumerate everything that is a bad idea
> because that list is very, very long.  Instead, we should stop asking
> people to do things that we think are bad ideas.  If there is a

I don't want to put everything up there. It's more like a FAQ list,  
but instead of questions there are rejected ideas. Magic symlinks come  
up from time to time. So it is a candidate for such a rejects-list.

> variation on the theme that could potentially be a good idea, but we've
> had several submissions that are not good ideas to date, then it's
> clear having it on the ideas list isn't helping matters.
>
>>> I have mixed feelings about "zombie" entries since we've reached   
>>> the point where most entries would be zombie entries.  How about   
>>> we have a separate page on projects that are currently in   
>>> progress?  People go to the ideas page, one presumes, to find   
>>> things to work on, so we should only list things that are new   
>>> ideas to be worked on.
>>
>> The metaphor behind my idea about the zombie entries can be   
>> visualized like as the plug-in window in firefox. It tells you the   
>> current status and when you click on update it will show te   
>> plug-ins which can be updated. When you update them the state   
>> changes in the list.
>>
>> Your proposal can be visualized as two tabs, one with the plugins   
>> for which updates are available (open ideas), and one for the   
>> plugins which will be activated at next (re)start (nearly finished   
>> ideas).
>>
>> For the firefox plugins the current way is more appropriate. For   
>> our ideas list I see good points in both approaches. I can't really  
>>  say one is more appropriate than the other. A variation of the   
>> zombie entries idea is to have a separate paragraph for the nearly   
>> finished stuff.
>>
>> My main motivation is to show the progess we make. Sometimes I get   
>> drive-by questions about the status of some of the entries. So our   
>> userbase definitivly wants to know about the progress. As long as   
>> we inform them instead of just removing the entries, It's ok for   
>> me. I don't care that much if this is inline, as a separate   
>> paragraph, or as a separate page.
>
> People go to the ideas page to find ideas for things to do.  Things
> already being done that don't need further help aren't things left to
> do.  We have at least one web page for in-progress projects -- how
> about we move the in-progress projects to that page and give it the

It's not the same scope. While each entry in the ideas list is a  
project on its own, it is most of the time not a big picture project  
like SMP or netperf.

The linuxulator entry on the ideas list for example is a project which  
would be suitable to put on the projects page. It is not there but we  
have some wiki pages about it. The wiki is more suitable for this  
instead of a projects page in CVS. There are contributors which are  
not committers which run for example the linux test project tests on  
amd64 hardware and update the status page in the wiki. We can add a  
link from the project page to the wiki pages, but I prefer a link from  
the ideas list to the wiki page. A link from the project page looks to  
me like there's a team working on this and they are proceding just  
fine, while a link from the ideas page actively shows that a project  
wants/needs help.

On the other hand, the ideas entry to extend dump/restore for better  
UFS2 support (backup of extended attributes) is not a big picture  
project which I would put up on the project page.

> same love and care that the ideas page is getting?  Or would you rather

Someone who is willing to put some love into the project page would be  
nice. There are several projects which seem to be dead. And if someone  
would want to continue with some of those projects he would need to  
restart from the beginning. So the page needs clearly a helping hand.  
Maybe we should put up an entry on the ideas list, maybe a volunteer  
is willing to ping all listed people about the status to determine if  
it is still worthwile to have all projects listed as they are ATM.

> we repurposed the current projects page to be a new ideas page, and
> pointed at that for new ideas instead?

In my eyes those 2 pages have different intentions while having a  
little bit of overlapping stuff. The stuff they have in common is not  
large enough to merge them.

Bye,
Alexander.

-- 
If at first you do succeed, try to hide your astonishment.

http://www.Leidinger.net    Alexander @ Leidinger.net: PGP ID = B0063FE7
http://www.FreeBSD.org       netchild @ FreeBSD.org  : PGP ID = 72077137