From: Mark S. <um...@no...> - 2002-10-20 00:39:58
|
On Sun, 2002-10-20 at 10:03, David Coulson wrote: > Mark Smith wrote: > > > > Actually, I'd argue the download page is the wrong way around. I don't > > think it is all that obvious, unless you already know where to look, in > > which case you have already had some "training". > > > > I think it takes a bit more than two seconds to find out. > > Really? Checking the date of your uml_utilities distribution against > that of the uml-patch isn't particularly difficult. For my project, having the latest networking code (within the stable kernel) is my primary interest. I've been compiling source to "upgrade" packages on the distribution I'm using for years. I don't bother with what version the distribution is using (well ok I do, but it's just a binary comparison - "does my distribution have the latest version ?" no, ok, download the best looking latest source (ie 2.4.19 patches rather than 2.5 patches in this case), ignore what's in the distribution). > > > Well, the message in that section doesn't mention that I need to > > download the right utilities. It does say "Again, if you go this route, > > you need to make sure that you have the tools you'll need to fully use > > UML.". Re-reading it, it is broad enough to cover the UML utilities, but > > it does not say that explicitly. As I'm about to compile a kernel, I'm > > thinking it is asking about having gcc/make etc installed. > > I believe that is detailed within the HOW-TO on the site. Not on the link that is pointed to in the "Building from source" though. It points directly to the page on compiling the kernel. If the HOWTO needs to be read from the start, all the way through, then specific links to specific pages should be removed from sections such as this. Having pointers to particular pages in the HOWTO tends to imply that they are stand alone instructions, that there are no information dependencies on previous pages in the HOWTO. > > > The change log in the source doesn't list say I need to make sure I have > > the latest UML utilities. And when I view the compile kernel page link, > > it doesn't say I need the latest UML utilities either. > > If you're building from source, using the development packages, you > really should subscribe to user-mode-linux-devel, as that is the only > mailing list which gets notification of new uml-patch and uml_utilities > development distributions. When a uml-patch requires a new release of > uml_utilities, Jeff will provide two Changelogs for the appropriate > packages which indicate that an update to another package is required. > Thusly, if you miss that, then you're not going to be aware that future > development releases will require the update. I shouldn't need to review the mailing list. I found a section which had a title implying that it contained everything I needed and needed to know, so I followed it. I did review the most recent 30 or so emails to user-mode-linux-devel mailing list, didn't see anyone complaining about the problem I was having. I didn't look at the developer mailing list, as I'm not a developer. I've been using linux since 1992, I suppose I still consider compiling from source code a user operation rather than a developer operation. Looks like I can add "linux developer" to my resume :-) I probably did consider reading the developer mailing list, but the instructions I found seemed to be so straight forward that my problem didn't seem to me to have been a "developer-level" type error. > > > If what I asked was an FAQ, it indicates that people are making the same > > mistake I did - it is not that obvious that you need to have the latest > > UML utilities. > > Most people don't know uml_utilities even exists when they go from > binary to source distributions. > > > Putting some extra comments in the relevant sections, or even better, > > moving all the source code required under the "Building from source" > > section would probably stop this FAQ being asked (and prevent a lot of > > end-user frustration - > > Indeed. > > > I've spent hours struggling with this problem - > > I've recompiled this kernel probably 5 to 10 times to try to fix this.). > > Sorry if I appear to be being overly pedantic about this, I just want to try to get some value out of the time spent and frustration I have had with this. Thanks for your help, Mark. |