Buchan Milne wrote:
Sorry, I was not aware it was this bad (DrakX not using vesa if it does not know the card).
It selected Vesa for me straight off, but seems not to have suggested it to the guy who had the problem. Don't know why.
BTW, you may have missed a lot of threads on cooker here where I have
been advocating that all display managers (not just gdm) kick off XFdrake if X fails to start ...
Didn't miss them. I agree with what you suggest, and thought this was a limitation of Mandrake that you would appreciate.
True. Mandrake has always been bleeding-edge.
Compared to say Redhat who ships with cvs snapshots of unstable versions of glibc (and then has to issue updates to glibc so users in large installations can see all users), and similar things?
I wasn't really comparing it - what matters is whether users should view Mandrake as a safe bet or a risky install. It seems from everything that has been said that they should view it as the former. But if we are comparing, Mandrake's origins are as an "upgraded" version of RedHat that included packages that RedHat left out because of licensing quibbles (e.g. KDE). So I guess you could say that it always used to be reasonable to expect Mandrake to be a little more cutting edge than RedHat. Whether that's still the case is another matter....
But if it's risky installing it before more experienced users have had a chance to test it, what were Mandrake doing providing it for inclusion with a UK magazine before it had even gone on public release?
Who said Mandrake provided it?
The accompanying eight-page article in the magazine - for one. This includes the comment that the screenshots were from RC2 as Mandrake provided the final packages too late for the publishing schedule. I take it from this that Mandrake cooperated in their inclusion. After all, as the public ISOs were not available at the time that the magazine was being prepared, I doubt Future Publishing (publishers of Linux Format) would have taken the risk of nicking the finished item from Club without Mandrake's permission. But maybe someone from Mandrake would like to clarify this.
And provides users who have already downloaded the ISOs (likely >85% of users) absolutely no incentive to pay Mandrakesoft, buy boxed sets, or join the club for no other reason but charity.
Exactly. I understand the logic of the decision to hold back the public ISOs for a while. But I don't understand the logic of taking this decision, and then undermining it by including it on the cover of magazines before public release.
Stability doesn't have that much to do with how many updates are available (IMHO).
Don't you have to reboot every time you apply a service pack? That doesn't help your uptime. And didn't the virtual memory management issues you mentioned cause stability problems?
Agreed, but you (and a lot of others) are concentrating on two hardware issues, and not seeing a lot of improvements in 9.2 (and the development process).
9.2 really is a substantially better release than 9.1, but no-one is prepared to tolerate even one hardware issue that affects them.
For my uses, the principal improvement in 9.2 is the inclusion of OOo v1.1. And they had to be badgered into doing this. You are a much more high-powered user than most, so you are probably better able than most to see the improvements, and I don't doubt that they're there, but what lesser mortals wanted above all was a release with as few bugs as possible, and that was yet again not the focus.
I think you are probably misjudging people's motives when they criticise the condition of 9.2 at release. I don't believe people are just trying to knock Mandrake. I expect most people want the best for Mandrake, the same as you. It's just that they are hoping to persuade Mandrake to change their priorities for the next release by pointing out how the current priorities (new features ahead of bug-fixing) cause problems.
And, I have seen many complaints on forums about beta2/rc1/rc2, and many of the reporters didn't report their issues, so how should they be fixed?
Of course, you have a valid point. But the volume of issues I have seen over the 9.2 final release is an order of magnitude greater than the issues reported to forums for the betas and release candidates. Simply because the final release is described as such and therefore seen as fit for public consumption. So it gets installed by a wider range of users, and those users don't expect the problems that one would naturally expect with betas.
Did you test the beta release? Did you report that your card wasn't automatically configured to work? Did you report that the vesa driver
worked?
I tested to the extent that I had time available and hardware to spare. That wasn't as much as I would have liked, but I run my own company and work a >80-hour week, so opportunities were few and far between. I would have tested the 9600 earlier, but I didn't actually own it until very late in the testing cycle. And it installed OK with the Vesa driver for me, so I wasn't aware there was an issue. And indeed, there may not be. This guy could have screwed up. That's not really the point. The point, as you yourself have pointed out, is that if installation fails, the user is left stranded.
Anyway, you are focussing on this one problem again. As I have tried to point out, this is just one of very many installation problems.
Mandrake can't test every single hardware configuration. That's why there are 4 or 5 opportunities for users to easily test installation,
to ensure the installation works on their hardware. If no-one tests
on the hardware you have, how is it to be fixed?
Buchan, from what I have gathered, you work in IT, and your job involves using Mandrake a lot. You are in a fortunate position to be able to test intensively. The majority of users aren't as lucky as that. To be able to test widely for hardware issues and user-friendliness, Mandrake need to include as many people as possible in the testing process. I, for one, cannot remotely keep up with their beta and RC release schedules, because I hardly get any time to play with this. I tried hard to keep up with the 9.0 and 9.1 testing phases, but (a) it was impossible, and (b) it was pointless (because the bugs I reported didn't get fixed - Radeon 7500 for 9.0, PDC20276 and DVB drivers for 9.1, for example). So I admit I didn't try as hard with 9.2 (particularly as I had even less time to spare). But I suspect most users will be nearer my situation than yours, and have limited time for testing (not to mention limited bandwidth for downloading every new ISO or updated package). So if Mandrake are fixed on a new release every seven or eight months, which naturally means the testing phase is compressed into little more than a couple of months, then they are never going to get widespread testing, because the average user won't have the time and resources to do it properly.
I agree that things must be improved, but saying "things must be improved" doesn't change the situation. Something needs to be done,
by those who are in a position to do something (ie those who have
affected hardware).
I believe most people would like to do their bit, within the constraints of their situation. But dismissive responses (by which I don't mean your replies to me, but earlier heated exchanges) to emails pointing out the various failings in 9.2 are not encouraging people to come forward and contribute. And Mandrake need to make it feasible for casual users to contribute to testing by allowing more time for the testing process.
Cheers,
Bruno
