httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Greg Stein <>
Subject Re: Apache 2.0 rant
Date Fri, 19 May 2000 09:37:36 GMT
I really hate to sound like an AOL-er, but I have to say "ditto" to Roy.
While I don't have much of substance to add to Roy's excellent response, I
*do* have to speak up to state that I emphatically agree with and stand
behind Roy's statements.

*) infighting? bunk. I haven't seen it. We pick each others' code apart
   and maybe we say "oh christ, that code sux." But we're talking about
   the code, not the people. Believe me: Ryan and I have butted heads a
   few times, even bordering on real harshness, but I think he's a great
   guy, and I easily dismiss the impact of those arguments when it comes
   to a *personal* interaction. Similar examples elsewhere. One example
   is that while I've been contributing very little code over the past
   year to Apache itself, I've been contributing a *lot* of review, 
   suggestions, and double-checking. And that is all that is -- it isn't
   calling Jim a bastard or Jeff a dumbass. It is code review, plain and
   simple. (give me a dozen beers and whip me with a wet noodle and I
   might call them that, but until then... :-)

*) the future of Apache? Quite rosy. Emphatic and blunt review of the
   changes means that we end up with *good* code. If punches are pulled,
   and statements are wishy-washy, then the "bad" code may never be fixed.
   Personally, I would love for somebody to tell me "dood. what the hell
   is that? you absolutely have to rewrite that function." That tells me a
   couple things: somebody has reviewed the code, and my code is not as
   good as it could be.

*) progress? It is forward. A lot or a little... it is *forward*. We have
   no deadlines, we have no schedule. All we have is positive, forward
   development. At some point, we say "rad. ship the sucker!" If the code
   looks hoky now... no biggy, we just ship it later. If it looks good
   now, then we ship now. "Future" is awfully relative when there is no
   hard and fast schedule.

*) API? trust to deploy? One word: "Feh." As with Ryan and Roy: if you
   don't like it, don't want to install it, and don't want to develop it,
   then what is the point? I can point to others who *do* want to do all
   that, and I'll give them the time of day whenever they want it. At some
   point, Apache will be "ready" and your opinion and choice will matter
   tremendously. I'd be quite happy to listen to your concerns. But to say
   right now that you're "worried", yet offer no assistance, then I'll
   just drop you into a particular bucket (exercise for the reader to
   describe that bucket). 

Hmm. That reads a bit harsh. Sorry, but I'm not sure how to soften it
properly, and I'm also not quite sure if it should be softened.


On Thu, 18 May 2000, Roy T. Fielding wrote:
> >Folks, this is what I was talking about earlier in the week.  I am starting
> >to become SERIOUSLY concerned about the future of Apache!  This infighting
> >has GOT to stop.  Now, yes, I have only recently started to read the mailing
> >list.  Basically about the time that 2.0a1 came out.  However, since then, I
> >have seen at least 3 big discussions about MPM and APR with what looked like
> >to me to end up with a bunch of bickering and egos flying around.  Now, I
> >may have misperceived this.  If I did, I apologize.
> You misperceived it.
> For all the new folks out there who haven't had time to read the past
> five years of Apache archives, you need to understand this: our recent
> discussions have been very mild compared to the churn and burn that
> accompanied the transition from Apache 0.7.3 to Shambhala to Apache 1.0.
> This project is run by mailing list, and the only way that people can
> express opinions is by arguing on this list, and even though that may
> seem less-than-cordial compared to the daily fondling of egos that
> occurs during traditional, face-to-face developer meetings, it does work.
> We just need to keep in mind that we are arguing about code and software
> design, not about each other.  I may argue til I'm blue in the fingers,
> but I am still more than happy to buy beer for anyone who contributes.
> We are in mass-QA mode right now.  It is absolutely necessary that everyone
> involved in development be able to express their opinions openly and
> absolutely without watering them down for public consumption.  This is
> where we prove that Apache is better software than that produced by 
> any single company behind closed doors and pleasant facades.
> >I will also grant that I have yet to actually look at the Apache code.
> >Ryan, I'm sorry.  Yes.  I am too busy but I also don't have a development
> >machine that I would trust Apache 2.0 on right now.  Further more, if there
> >is LITTLE or NO documentation of the API, then I do not see this project
> >moving forward very far.  :(
> Dude, on this part I agree with Ryan: If you aren't contributing to the
> development or testing of the product, then why should we care if you are
> worried about it?  We aren't doing this work for the faceless millions
> who use our products; we are doing it because we need web server software
> of the highest possible quality for our own projects, our own companies,
> and our own communities.  If for some reason we are failing to support
> some community of users, it is necessary for them to join in supporting
> the server project, not the other way around.  Because that's the only
> way you get code that contains the features that people actually need.
> I am not at all worried about the progress of 2.0 development, especially
> seeing it grow exponentially over the past two months.  Some of the stuff
> that Bill+Bill have accomplished over that period has been more significant
> to the win32 port than anything since 1.3.0 was released.  I love this pace.
> Apache development is non-linear.  I've seen people here produce higher
> quality code in one nights work than I have seen from Microsoft in an entire
> year's worth of effort by 400 programmers.  Quantity doesn't matter.
> Schedules don't matter either.  What matters is the quality of the final,
> non-alpha, non-beta, release.  That is what will make 2.0.0 better than
> the most stable 1.3.x.
> ....Roy

Greg Stein,

View raw message