1 |
Mark Knecht posted on Thu, 27 Aug 2009 07:56:25 -0700 as excerpted: |
2 |
|
3 |
> We need better tools for creating and maintaining personal overlays. |
4 |
|
5 |
I'd like to explore that comment a bit, but below... |
6 |
|
7 |
> Here is my story. |
8 |
> |
9 |
> Support for old hardware has been one of the downfalls of the devs |
10 |
> deciding to not keep everything in portage but rather they start weeding |
11 |
> out software before we users have really finished using it. I have 4 |
12 |
> [old but special machines I use] as MythTV frontends. |
13 |
|
14 |
> At the time I bought the machines the Open Source radeon driver didn't |
15 |
> support S-Video so I had to use the ATI driver which worked fine. |
16 |
|
17 |
When I first switched to Linux, I was in much the same boat. While I had |
18 |
been researching a switch for a couple years, and had therefore verified |
19 |
that all my hardware purchases "did Linux", I wasn't yet aware of the |
20 |
difference between "Linux drivers" and "freedomware Linux driver". I had |
21 |
thus verified that my nVidia card with "Twinview", one of the reasons I |
22 |
purchased it, did "Twinview" on Linux as well, but unfortunately didn't |
23 |
know to verify that the FLOSS drivers handled it. Of course, the open |
24 |
source nv driver doesn't, due to lack of vendor cooperation. |
25 |
|
26 |
Shortly after I actually /did/ switched to Linux and discovered all that, |
27 |
I vowed that while I couldn't at the time do anything about the video |
28 |
card I was running (no budget for a new one), that was the LAST time I |
29 |
would make that mistake! |
30 |
|
31 |
And it was the last time I DID. My cards since then have all been |
32 |
Radeons, and I've only purchased ones well supported by the freedomware |
33 |
drivers, which is why I'm still on a 9200, as that was the best |
34 |
freedomware supported for quite awhile, until AMD bought ATI and turned |
35 |
them around in that regard. |
36 |
|
37 |
FWIW, I expect to upgrade video again shortly, I've had it penciled in |
38 |
for 2H2009 for awhile and that's what it is now. It'll be to another |
39 |
Radeon of course, as with AMD at the helm, that is again the best |
40 |
freedomware choice. As I'm still AGP, and the hdXXXX series (r6XX+) do |
41 |
better on PCI-E, I'll probably get the top of the line r5XX series, the |
42 |
x1950, altho I'd much prefer $100 to the $150 it still seems to be |
43 |
running. |
44 |
|
45 |
> A couple of years ago ATI, in all wisdom, dropped TV Out support for |
46 |
> this specific chipset from their closed source driver so I was forced to |
47 |
> stick with the driver that was current at that time. |
48 |
|
49 |
Yeah. Well, as you note below, at least the freedomware drivers are |
50 |
picking up support for it now. |
51 |
|
52 |
> This was OK for awhile as it was in portage and I could just mask |
53 |
> higher revisions. However after awhile it turned out kernel updates |
54 |
> forced incompatibilities between new kernels and this old driver so I |
55 |
> was forced to mask newer revisions than the last one that worked with |
56 |
> the last radeon driver that worked. |
57 |
|
58 |
Yeah, that happens. Expecially when you're running servantware drivers |
59 |
for hardware they've quit supporting. |
60 |
|
61 |
> 3 months go by and portage maintainers decide to start weeding 'old' |
62 |
> software out and, you guessed it, they weeded out what I needed to run |
63 |
> this hardware. |
64 |
|
65 |
Just a minor term correction. As the term is normally used, "portage |
66 |
maintainers" would refer to those maintaining the portage package. |
67 |
|
68 |
What you want would be something like "Gentoo devs" or "Gentoo tree |
69 |
maintainers", or here, it would have been the "package maintainers" for |
70 |
the packages in question, since newer versions of the same package |
71 |
remain. If the entire package, all versions, is removed, that's often |
72 |
the tree cleaners project, or the devs in charge of the herd the package |
73 |
belonged to. But in that case, there's an established procedure of a 30 |
74 |
day package-mask and an announcement on the gentoo-dev list, in case |
75 |
another dev wishes to rescue it, and to give any users that may still be |
76 |
using it time to react, putting it in their overlay or switching to a |
77 |
different alternative, before the package is actually removed. Plus, by |
78 |
the time it's removed, it's often dead upstream, abandoned, no longer |
79 |
compiling with current gcc and/or against current glibc, open bugs with |
80 |
little chance of fixing them, a Gentoo-dev maintainer that's either long |
81 |
retired or no longer interested in maintaining the package, etc. |
82 |
|
83 |
Of course, this whole thread is about how kde3 is heading there, but it |
84 |
wouldn't count in this regard since there are newer versions, so it's not |
85 |
entire package removal, except for the kde3 packages that don't have kde4 |
86 |
versions, in which case, yes, they'll go thru the mask and announcement |
87 |
for removal process, before final removal. But one of the big |
88 |
differences here vs normal upgrades is that there's a real possibility |
89 |
they'll end up masking kde3 before kde4 goes stable, according to the kde |
90 |
meeting summary. Plus, all the various things still keeping people on |
91 |
kde3, some of which are why kde4 is /not/ yet stable. |
92 |
|
93 |
Meanwhile, you didn't do it here, but in general, and it's a habit I've |
94 |
had to correct myself as well, also note that what used to be termed the |
95 |
"portage tree", that is, the main Gentoo tree, really shouldn't be |
96 |
referred to as the portage tree anymore either, as there are other |
97 |
package managers (PMs). For historical reasons, it's also often called |
98 |
the gentoo/x86 tree, even tho that isn't really accurate either since |
99 |
it's generally scripts for building from source on all archs (as we know |
100 |
in the context of this list). But I guess the original Gentoo folks |
101 |
didn't realize that, and created their tree under an x86 subdir. |
102 |
|
103 |
The most accurate way to refer to the main Gentoo tree, therefore, would |
104 |
be just that, or simply the Gentoo tree, or the Gentoo repository, since |
105 |
that's it's official name now, as can be seen in profiles/repo_name, |
106 |
distinguishing it from other repositories/overlays/trees, including those |
107 |
of Gentoo based distributions such as funtoo. |
108 |
|
109 |
> The ATI driver was gone. The kernel was gone. No |
110 |
> discussions, no announcements. It was just gone. |
111 |
|
112 |
As mentioned, it's upto the package maintainer what versions he keeps in |
113 |
the tree, and there's no special process necessary when he decides to get |
114 |
rid of one, except that ordinarily, they're not supposed to remove the |
115 |
highest keyworded (either ~arch or stable) version without consulting |
116 |
with the arch team in question. Of course, that's part of the process |
117 |
the kde team is going thru now, warning other devs that kde3 may end up |
118 |
masked... |
119 |
|
120 |
Only if a package is removed entirely does the package removal process |
121 |
trigger, masked for 30 days with an announcement on -dev, then a removal |
122 |
if no-one has stepped up to maintain it and no other serious overrides. |
123 |
|
124 |
> A machine I could build and run using a Gentoo 2006 install CD could no |
125 |
> longer be built and run using a 2008 CD. |
126 |
> |
127 |
> Then I'm forced to learn about attics, building overlays, etc. It was a |
128 |
> mess for a long time. |
129 |
> |
130 |
> Recently the Open Source driver has started to support TVOut on this |
131 |
> version of the Radeon hardware, so I'm now back to using Open Source, |
132 |
> but video quality is FAR inferior to the ATI driver, although CPU usage |
133 |
> is far superior so at least with OS I have a quiet machine while |
134 |
> watching a bad picture. |
135 |
|
136 |
Well, might be poor solace, but at least you can see the good with the |
137 |
bad. |
138 |
|
139 |
> Moral of the story - don't trust portage to support your machine |
140 |
> tomorrow just because it works today, and don't expect portage |
141 |
> maintainers to care. The response you'll get, if you get one at all, is |
142 |
> 'be a man, create your own overlay, be responsible for your machine, and |
143 |
> shut up'. |
144 |
|
145 |
Umm... package maintainers, but that's covered above... |
146 |
|
147 |
Of course, I'd say another moral is to pick hardware that's FLOSS |
148 |
supported from the beginning, if at all possible. Sometimes it may not |
149 |
be, and then one has to decide whether it's worth the negatives to go |
150 |
servantware or not. It's a decision I'd make differently than many here, |
151 |
but it /is/ an individual decision. |
152 |
|
153 |
But, being able to create overlays and the like is one of the features of |
154 |
Gentoo, and I'd agree with them in that regard. However, just because |
155 |
it's true, doesn't mean they shouldn't at least provide some pointers to |
156 |
documentation on creating an overlay, etc. After all, it's other humans |
157 |
with very real problems of their own they're dealing with, and it's good |
158 |
to remember that just as us users need to remember that when we wonder |
159 |
why there's no apparent action on our bug. |
160 |
|
161 |
But that brings us back full circle to what I mentioned above. As |
162 |
someone who has gone thru it as an ordinary user, what /did/ you find |
163 |
most difficult about the process of creating your own overlay? What sort |
164 |
of tools are you referring to, that could make the process easier? I'm |
165 |
just a user too, but am perhaps more technical than most, and spend |
166 |
enough time on the dev list and etc, that while I didn't find the process |
167 |
particularly difficult, and don't really understand what sort of tools |
168 |
could have made it much simpler than I found it -- it was just a matter |
169 |
of taking the time to do it -- I can't take that as indicative of the |
170 |
problems an ordinary user may have. |
171 |
|
172 |
The reason I'm asking, is that it's just this sort of feedback that devs |
173 |
often lack, and if between us and others here who may want to speak up as |
174 |
well, if it can be hashed out exactly what is causing the problems, |
175 |
perhaps I can present them to somebody who can do something about it. |
176 |
After all, it's not like the devs generally go out of their way to make |
177 |
things more difficult. If there's a reasonable way to make the process |
178 |
easier, and it's not going to take an unreasonable amount of programming |
179 |
to make it so, there's a reasonable chance something can be done about it. |
180 |
|
181 |
And... I'd not mind being a part of the project and any solution that |
182 |
might come of it. After all, not only might I find it's easier for me |
183 |
too, but that's a another way I may be able to make my own contribution |
184 |
to this great big community project we call Gentoo, the even bigger one |
185 |
that's Linux, and the even BIGGER one that's the FLOSS community and |
186 |
projects in general. I may not be a coder, but if I can do something to |
187 |
help, I'll certainly give it a try! =:^) |
188 |
|
189 |
And of course the same applies to you. If any changes come out of this, |
190 |
you can point to them and say it's because of you! That's a very nice |
191 |
feeling to have! =:^) |
192 |
|
193 |
(Of course, it's a similar feeling to the one I get as a reward for all |
194 |
the work I put into my posts here, when I see folks saying they actually |
195 |
find them useful, even to the point of archiving them. =:^) I remember |
196 |
the first time someone mentioned saving my posts, back a very long time |
197 |
ago both in years and in changed personal situation, as it was about a |
198 |
decade ago, and it was a post I made to the IE4/OE4 beta newsgroups... |
199 |
how times have changed indeed, but you should have seen me after reading |
200 |
that comment, as I must have been walking about two feet off the ground! |
201 |
I know because it took me about a week before my feet touched ground |
202 |
again! =:^) That's a VERY powerful reward indeed! =:^) |
203 |
|
204 |
-- |
205 |
Duncan - List replies preferred. No HTML msgs. |
206 |
"Every nonfree program has a lord, a master -- |
207 |
and if you use the program, he is your master." Richard Stallman |