Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm List-Subscribe: List-Archive: List-Post: List-Help: , Sender: cygwin-owner AT sources DOT redhat DOT com Delivered-To: mailing list cygwin AT sources DOT redhat DOT com Message-ID: <3B9BC245.7010805@ece.gatech.edu> Date: Sun, 09 Sep 2001 15:25:57 -0400 From: Charles Wilson User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.2) Gecko/20010713 X-Accept-Language: en-us MIME-Version: 1.0 To: Greg Fodor CC: cygwin AT cygwin DOT com Subject: Re: Windows XP Fix References: <000001c13958$95e9fe50$9801a8c0 AT fuzzy> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Greg Fodor wrote: > No need to get nasty. Christ. Sure, Chris went a little over the top. But, it was *completely* justified -- put yourself in the shoes of the core developers. Take a look at the furball that's erupted over the last week with cygwin-1.3.3(no, not ready yet). We've all got "real" jobs; this is volunteer work -- and there's plenty to keep all of us busy 24 hours a day on CURRENTLY RELEASED OS's. It stuns me how many people are doing beta testing for Microsoft's latest closed source nightmare, but don't want to dig in to the problems they encounter with open source projects on that platform. Yet they expect us -- who do not have this AS YET UNRELEASED operating system -- to fix something in the code for them. > > I spent like 3 hours screwing around with my cygwin install and surfing > archives Then surely you saw how many people continually report that "sigwin doesn't work on XP" or "cyguin works for me on XP". The fact is, *we do not care*. Yet. And in three hours, you could have learned a lot more about the cygwin code itself, and perhaps contributed to a solution -- instead of coming up with, as Chris puts it, "a magic incantation". > in order to get it working under XP for the time being until > someone fixes cygwin1.dll. "someone"? Who would you suggest? Would you like to nominate me? Or Chris? or Egor? I'd imagine you'd have to talk to our "real" bosses if you want to schedule our time for us. Oh, yeah -- and buy us all copies of XP once it hits the streets. (BTW: I sure as HELL am not going to use XP -- ever, if I can help it. The more I read about it, the less I like. I'll stick with 2000 until I finally and completely switch over to linux for desktop work.) I'm pretty sure Egor, Chris, Robert, Corinna, etc are NOT all panting and bursting with anticipation to rush out to Fry's at midnight to pick up a copy of XP as soon as it's released. Point; Free software works like this: programmers scratching an itch in their free time. *You* have the itch. Go scratch. > I offered my hacked solution, something I did > to get _my work done_ as a temporary way of working around the problem > until the correct solution is found. I didn't investigate into the > September 1st cygwin1.dll snap yet, though I'm guessing that's the > optimal solution. Wait -- so cygwin was in your critical path for getting work done. Ergo, you had a working cygwin install at some point -- running on some released version of windows (9x/Me/NT/2K). Then, you installed a BETA version of Windows -- without checking first to see if cygwin worked on it -- and you did this on your primary work machine? So the breakage in your critical work path is OUR fault? Umm...no. It's your fault. *NEVER* install a beta OS in your crit work path, unless your job is writing programs and/or drivers that must hit the street when the OS is released. (But if that were the case, then you wouldn't be using cygwin at all -- so I doubt that). > > Regardless, you should offer your users *something* on the webpage. The > "sorry, come back later" message there now, frankly, blows, since more > information is available for people who want to spend the extra effort > to get their stuff working. Sure. If somebody would provide REAL information. As in, "there's a bug in the malloc implementation of cygwin that causes problems on XP. This will be fixed once XP is released and some XP-users can join the development team, but for now, the cygwin snapshot from Sept 1 seems to be okay. HOWEVER, XP is NOT officially supported yet, so do not be alarmed when you get some grumpy responses if you complain about problems on that platform prior to XP's official release." Not "replace your bash.exe and magically your tcsh will start working if you've installed the cygwin-1.1.2 dll." or whatever it was that you were suggesting (I know, that "quote" is an exaggeration...) > The fact that some people have gotten XP > working by changing the DLL, or by switching bash.exe, is enough to > warrant a quick-fix note on the page to avoid the hassle of forcing > users to figure it out on their own. Don't you care that people are > pissed off right now because cygwin is broken on XP? No. I don't care. Remember, free software is about freedom: I'm free to scratch MY itch by digging into the code. My itch collection does not currently (and will not ever, if I can help it) include "XP". YOU are free to scratch your itch, which apparently has XP very high on the list. But not high enough for you to actually download, study, and debug the code on your UNRELEASED platform. > (Which BTW, is in > the hands of OEMs and being installed on PCs as I write this, AFAIK.) If > you are waiting for the retail release, people who have XP on their > computers now from OEMs and beta people will be forced to wait another > month and have broken systems. fine. So buy Corinna a new computer with XP preinstalled. (Be sure to get the appropriate export licenses for shipping a computer and software overseas). Or buy Chris a new computer. The fact is, cygwin is NOT part of anybody's job description. *It is volunteer work* -- even for Chris. [okay, corinna may have cygwin in her j.d. -- I'm not sure.] Geez -- I'm already a year overdue with my Ph.D. thesis; you want me to spend MORE of my time on cygwin so that it works on an OS that I personally don't give a rat's **** for? I'm not going to go out and buy a new computer with an OS that sounds like worse *@^& that usual, in order to make cygwin work better for people who don't have the gumption to help themselves. Nor am I going to go out and spend $$$ on this year's "revolution" from Redmond. I *do* real work on my primary machine; it's happy with 2000. I ain't changin' nuthin' until my thesis is done. Frell! Why can't those of you who HAVE the damn OS do some #&!@##$ work, fer gossakes? and those people who DO buy a new machine with a preinstalled XP -- do you think that some of THEM might be programmers? Do you think that some of THEM might share your itch -- but also some initiative? Sorry. lost it there for a minute -- but this "topic" really irks me. > I realize that it doesn't make any sense as to why running programs from > within an old copy of bash keeps the problem from occurring, but the > fact is it does and I can now run my PostgreSQL server within bash on XP > whereas I normally couldn't. > > The point is, nobody gives a damn about the reason something is broken, > simply enough, it's broken. If my sink is leaking all over my floor and > all I have is the piece of gum in my mouth to fix it for the time being, > then, fuck yes, I'll use the gum until the plumber gets there. Look. If you want cygwin to "work" on XP, then *developers* will have to understand WHY it doesn't work. Otherwise, we can't fix it. Magic incantations are just that -- magic. It may be a race condition that appears on *your* version of XP -- and is somehow fixed by variant execution delay paths in bash-2.0.5-2 instead of -7. On your system. With your collection of background services. And your optimization tweaks. On your filesystem. And with your particular collection of XP core dlls -- which may be different from the final "gold" ones. Since it's just a magic incantation -- there's no guarantee, and even less likelihood -- that it will fix the problem on someone else's betaXP system. What you have is an anecdotal data point. But without understanding, there's no way to extrapolate that to even "workaround" status. There's no reason, then, to post a list of anecdotes on the webpage; that's what the mailing list is for. (Well, that and complaints about brokenness coupled with personal excuses). Oops. Lost it again. > > I subscribed to this list in order to post what I had found so as to > help others, Your intent was good. Your message (and esp. followup) was like a knife on raw nerves. We've been over this and over this on the list. *WE* don't have XP. Therefore *we* can't fix it. Therefore, *somebody else* will have to fix it -- at least until/unless a core developer takes the XP plunge. And suggesting magic "workarounds" *impedes* progress toward getting someone who actually HAS XP to actually FIX the bug. It's because magic incantation/workarounds actually *IMPEDE* fixing the bug, that suggestions and presentations of those "workarounds" get flamed so hard. We want cygwin to work -- not kinda sorta work if you stand on one leg and quote the Veda in its original sanskrit with incense burning. (this is not ridicule; this is exaggeration for humorous effect) Therefore, we'd like for XP users (at least a few of them) to take *current* cygwin software and debug it -- not encourage other XP users to stay off the mainline of developent and use old versions that kinda sorta work. > and I get ridiculed. Look, mailing lists are not tea rooms. This one is actually pretty mild. (For kicks, go to the linux-kernel mailing list and suggest a C++ rewrite. Or that Linus use CVS to track kernel patches. Or that the linux-developers accept some patch to work around a bug in a (closed-source, non-free) NVidia graphics driver. But put your asbestos suit on first...all of those suggestions are things that (a) have been discussed -- acrimoniously -- many many times before, (b) REALLY get under the skin of the core developers, and (c) have been authoritatively, definitively, and finally answered by Linus himself. Kinda like cygwin-on-XP here :-) I lurked on the cygwin mailing list for *SIX MONTHS* before I made my first post. (yes, I walked to school 10 miles uphill both ways barefoot thru the snow in the summer.) It took that long until I was sure I understood the background, culture, and project goals so that I wouldn't (a) embarass myself or (b) step on a land mine (XP...) and get flamed. > Very nice of you, I'll be polite and > assume you don't represent the rest of your peers. Well...I don't like to flame people, but I'll do it if pushed. Chris has been pushed, lately, so I cut him some slack on that point. However, the attitude of "we don't support beta OS's until they are released"? Yeah, I think most of us agree with that. As I said, we've got enough to keep us busy for months -- on currently released OS's. Now, I just spent about an hour answering your email. That hour could probably have been better spent on (a) cygwin development (b) my thesis (c) watching TV (d) sitting out in the sun on my porch. That's another reason contentious topics get flamed -- we don't like re-answering the same question over and over -- especially those that require lots of explanation, like yours. But, we do it anyway -- because hopefully, through the flickering flames, somebody somewhere might actually learn something. A truly unfriendly list would just bit-bucket everything it disagreed with. --Chuck -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/