Hi, I like the foxboard very much and also appreciate the work by John and everybody at ACME...
But I also understand what upsets Riccardo: What he suffered from is simply that known facts (problems) are not passed on to users and (potential) customers. Take for instance this post I send in january: >Hi all, >in the realease notes for the current SDK it says : USB support is unstable >in kernel 2.6. Is there any more detailed information on that? >Is USB support broken at the core-level or just for special kinds of >devices... ? >I would like to port a driver for a new device, that uses control-tranfers >and interrupt-transfers (read/write). >Will I get in trouble with these basic requirements. >Thanks >Eberhard Fahle As I know now (from the current discussion) that problem can be tracked down to a "heavy load" thing, and it seems to be known for quite some time. Why did nobody care to tell me that? That would have been a one-liner ... -> "It's a heavy load problem" Since I only send a few bytes over the USB-Port, I wouldn't have cared. With some more information I would even have tried to help tracking down the whole issue for a solution. In this case you are ignoring potential help from the community. When I had a look at the update of webpage for the WIFI-device http://www.acmesystems.it/?id=705 I way actually annoyed by the fact there is a troubleshooting section now, but please ... why at the very end of the page? I don't opt for starting the page with problems, but a line reading : "This is about the WIFI thing, works fine under normal load (for heavy users see section troubleshooting)" would clear things up. How can situations like these be avoided in the future ? Riccardo Cassinis wrote: > one reads technical specifications, not mailing lists, when deciding which > device to buy He is right! But I _always_ have a look at a forum for every product I buy. I visted a forum for tumble dryers last week! There where 3 years worth of tumble-dryer information on this site! Information that is available on a mailing list is lost for new users. One of the reasons why the same questions reoccur all the time. In a forum you have easy access to every topic discussed in the past. Usually it's nicely grouped into sections Hardware,Software, Networking... If that forum has a "know bugs section" this discussion would probably never have happened because Ricardo would have searched for "WIFI" before buying and without registering with a mailing list. Mailing lists are for geeks, great for passing around code and patches, but not information. This maybe sounds too simple, but think about it... Eberhard On Friday 15 June 2007 10:45:45 > John, > > first of all let me thank you for the time you took in writing such a > detailed answer. > > Let me point out that in my message I never blamed, nor even mentioned > your work, which I appreciate very much. I fully support open software, > and I know very well the amount of efforts developers devote to it. > > As for the point wether communication through the wi-fi channel works or > not, from my point of view as software tester and user I would say no, > it doesn't, because the system should be capable of understanding when > the critical limit is about to be reached, or should provide a means for > doing that. I can't obviously expect full bandwidth from a 100 Mips > machine, but I don't expect that the whole system unrecoverably crashes > either! Anyway, it should be clear that I am NOT blaming on you because > the software does not yet work as it should (I am confident that the > problem will be solved soon). > > You state that the Foxboard runs Linux, which carries no warranty for > any particular purpose. This is true. But the Foxboard is not software, > is not free, and is advertised as being fit for a particular purpose. > This is what upsets me. Saying that the problem was well known since > long and that it had been discussed in the maiiling lists does not help > in this sense. Usually, . Specially if the manufacturer > has a good reputation (what i meant with my sentence "doesn't add much, > so to speak, to Acme's reputation" was that I was surprised that a > company that had a very good reputation would advertise something > untrue). The recent addition to the web page proves that I was right, > although the definition of "normal" and "heavy" load is vague, and a > lawyer would surely find reasons to complain (once again, with the > hardware manufacturer). > > Ok, we shall now place that project "on hold" waiting for better times > to come. Let's see what the other people who have the same problem will > do. But if anyone finds a workaround, please let us know! > > Riccardo (double C, please...) > > --- In [email protected], John Crispin <[EMAIL PROTECTED]> wrote: > > * i have taken a long time to think about what to say and how to say > > it so please everyone read it fully. If anyone would like to comment > > on the matter you are invited to do so, from the bottom of my heart * > > > > dear ricardo, > > > > i am john crispin, the maintainer of the sdk toolchain. When i started > > using the foxboard, when they were first released, the sdk came > > straight from Axis in sweden (the producer of the chip). it was > > missing a lot of drivers (like wifi) and packages (like wireless > > tools), that i added to the sdk using what was previously know as the > > "phrozen patch" and is now known as the FoxSDK. Being an OpenSource > > developer, i did this fully in my spare time, contributing to the > > community of nice people, just for the fun and because i have the > > knowledge to do so. All the code in the sdk is under the GPLv1/v2 > > license with the clauses probably known to you (but we will come back > > to this later in the mail). > > > > *** hence this statement is purely personal and not an official > > Acmesystems statement. They will contact you directly very soon *** > > > > Let me respond now to the interesting mail that you send to the list > > > > > Being deeply involved in Marco's and Cristian's work (I am their > > > supervisor), I regret to say that the whole matter is being treated > > in a > > > > completely wrong way. The problem is not wether kernel 2.4 is more > > or > > > > less stable than 2.6 as far as wireless is concerned: the point is > > that, > > > > on Acme's website, the Foxboard is advertised as WORKING with wi-fi, > > and > > > > specifically with that particular USB stick (the DWL G122). > > > We therefore bought the device FROM THEM trusting the ad, and later > > > discovering that > > > > You are correct, that on the acme site it is stated that the stick is > > supported. Form my point of view as a developer it does work. There > > are known issues that the 2.6.15 support of the usb host driver > > provided by the cpu manufacturer has problems under heavy load. I did > > respond very quickly to the bug report issued by your students, > > however i regret to haver sent only a short mail. The reason for this > > is that the issue has been reported lots and lots and lots on this > > list and it has also been announced just as often that a fix is being > > worked on. As i said i do the developement of the SDK in my free time > > and under GPL, so some times it takes a bit longer to have a fix fully > > working and tested. I do however agree with you that the information > > on the website is not complete, it states that the wifi stick works, > > however with the limitations i explained above. > > > > >>>>> yes this is a known problem > > > > > > and that > > > > > >>>>> unfortunatley the issue will not be resolved until we > > >>>>> have 2.6.19 > > > > As i said i am working on a fix, there will be a lot of new software > > using a new release, which is a complete rework of the code base, > > adding more new features than most would dream of. With the upcoming > > release i will prove once again that the community means alot to me > > personally, otherwise i would not invest hundreds of "free" hours into > > the support. I am in constant contact with a lot of users, reachable > > via skype, icq and even telphone, giving free support where i can, but > > i am not a magician and 100 mips is only 100 mips. making usb stable > > on 2.4 was hard work and so will the part of making it stable on 2.6. > > However as i said it does take its time. > > > > > At this point, I think there are only two possibilities: either, as > > the > > > > Italian law states, the problem is fully solved (by Acme, not by us) > > > within a REASONABLE amount of time, or they will have to provide a > > full > > > > refund for all expenses we incurred, given that their product is not > > > conforming to advertised specifications. And this means that we need > > a > > > > VERY FAST response from Acme. > > > > I am from germany, so i have no idea of the italian law. But what do > > you understand as a very fast response and reasonable amount of time ? > > i spoke to Mr. Asquini - Co-Founder of Acmesystems SRL. today to > > discuss your e-mail, he will contact you directly on friday 15.june. > > as i stated before, in my opinion the producer is "conforming to > > advertised specifications" just that there are known issues when using > > webcam+wifi > > > > > I am sorry for being somewhat rude, but we are all wasting precious > > > research time on a trivial problem that someone else claimed did not > > > exist. > > > > dont worry, i am sure you took as much time as me to think about the > > words said. And rest assured you REALLY said them clear enough. > > > > now for the really shoking part > > > > > doesn't add much, so to speak, to Acme's reputation. > > > > I have meet the guys from acme on several occasions and i have only > > rarely meet such nice supportive people. I have worked with linux > > hardware for many years now and the foxboard is by far the most fun i > > ever had with any embedded device. It has been taken from the whole > > linux community with a lot of anticipation and joy. I have seen lots > > and lots of great project using the fox, from industry automation, > > train controls, roborts, toy helicopters, routers, print servers, ... > > (i have not enough time to give you a full list, but believe me it is > > long). Up to know you are the first person i have come across that is > > so unhappy with his fox in the 2 years that i have been using it. > > especially as the problem is well documented in the mailing list (and > > yes i know it should be noted on the website, which was added just 2 > > minutes ago). So basically your statement implying Acme has a bad > > reputation is shrudely false. I grant you your own opinion however, i > > just believe it is totally none representative from an academic point > > of view. > > > > > > Finally i would like to point out to you the following. The device you > > are trying to use runs linux. By using the software you implictly > > accept the GPL which clearly states the following text > > > > --- quote --- > > This program is distributed in the hope that it will be useful, > > but WITHOUT ANY WARRANTY; without even the implied warranty of > > MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the > > GNU General Public License for more details. > > --- quote --- > > > > thanks, > > John
