Hey Harv, apparently the moderators haven't had a chance to notice that this is one of your usual reposts to all *BSD threads. That's the only way this post could still be labeled 'interesting'.
Well, Harv, I don't see how your reply has anything to do with the statistics I posted. Nevertheless...
> the FreeBSD project has only gotten more fractious and hateful in recent memory
You must not read the same mailing lists I do (most of the development ones). The last round of major flame wars died out around March/April and things have been much smoother since then.
Of course I don't expect this to influence you, as you're clearly only interested in repeating yourself without taking in new information.
> but the arrogance and condesendence of the *BSD community (and to be
> fair, it exists, to a much lesser extent, in the Linux community as well)
> has turned me off [...]
Well, I'll finesse the issue of trying to decide which community has more of this bad attitude by saying, why can't we just say this attitude isn't constructive, from anyone, period, and then just move beyond it?
My own personal opinion is that most people doing _real_ work on any of these projects are too busy to play the zealotry game. There's too much stuff to do. (In fact, I think I'll shut up on this thread now and go back to bufixes myself.)
That's right, msmith left because he didn't consider it fun anymore. Others have left too. Others stayed. Others joined. So what? that's the nature of volunteer projects. Why rehash this posting again and again?
It would be nice if you would attribute the original author of this email. IIRC it is from about a year ago. Some people left, some people stayed, life went on, as in any other volunteer project.
Boy, it looks like the parent post got under your skin much more than any other recent *BSD post I've seen. Instead of just posting your standard messages once per thread, you've posted each of them, what, 3 or 4 times in this thread already?
I ask again, (as I asked earlier but not under this userid -- I have been too unmotivated to create a slashdot userid before), what is your goal in these repeated postings? Why not just avoid using the damned thing if you hate it? Why the repeated trolling with the exact same postings over and over again, in multiple threads, with differing subject lines?
> In the FreeBSD Ports Collection, there are many Ports marked as broken,
> and many more unmaintained and suffering from bit-rot.
Let me inject some facts into this discussion, as this is an area in which I've done a lot of research.
As of 12/30/2003, there are 10011 ports in the FreeBSD ports collection. Except for a very few cases where only a binary is available from the authors, the entire collection is meant to be buildable from source.
On 4-STABLE i386, 214 ports fail to build from source; on 5-CURRENT, 366 fail to build. The difference is primarily due to FreeBSD being an early adopter of gcc3.3 in 5-CURRENT. As these bugs are fixed, we try to encourage people to get them adopted upstream. This benefits the Linux as well as BSD communities.
239 ports are marked broken on 5-CURRENT; these are the 'hard-broken' cases. Right around 50% of those are compile problems and in some cases the code is indeed quite old. Some of the others that fail to build, but are *not* marked broken, are failing to fetch and part of that is due to the savannah compromise which is outside our control. That's temporary, we all hope. (I'm reluctant to see any port marked as broken unless it really won't compile/install -- not just having some transitory error.)
A slightly larger number, for which I don't have statistics, are marked broken on the 64-bit architectures, generally due to bad assumptions in their C code. Again, as these are fixed, changes make it back to the larger community. And, the number of broken ports on the 3 64-bit architectures has also come down dramatically.
The number of ports PRs has come down from a peak of 1500 during the 5.1-RELEASE freeze, to around 800 today. This is primarily due to a dozen new ports committers in the past 6 months who have been very active.
The number of officially unmaintained ports is 2616. Although this number may have decreased as well, clearly we still need more volunteers.
As for bit-rot, again, much of the code that is rotting lies outside the control of the FreeBSD ports team. Another project that is active is to try to identify, and prune, ports for which there is no hope. However, this requires getting a community consensus on what "no hope" is, and that takes time. The last time a pass was made scheduling ports for demolition, somewhere around 100 were proposed, and around 70 were fixed by someone or otherwise adopted in the 3 months afterwards. There are another 100 or so proposed for removal in the next pass; some of those have already been fixed as well. (I hope to create a framework for making this process more visible to end-users so there are as few "surprises" as humanly possible.)
One last point about bit-rot. For ports that fail to compile/install, the previous version of the binary ("package" in FreeBSD terminology) remains available for fetching until the new version works. Thus, most of the bit-rot only affects those who are installing from source.
The code that I've written that generates these statistics does so by mining the Problem Report database, the error logs from the bento build cluster (which iterates over each port on each architecture and each relevant OS release, continuously), the ports collection itself, and (to a limited degree) CVS meta-info. As far as I know no-one else has similar functionality. (Interested parties can contact me directly; since this machine is on the end of a cable modem line, I need to protect it from slashdotting). I hope to get this code up and running on a machine with greater bandwidth to make it more generally useful, soon.
To conclude, IMHO things that get fixed in the FreeBSD ports collection can have the effect of helping the greater community. I'd like to see the culture of zealotry start to disappear, and a greater degree of peer respect, especially in regards to making as many apps useful to as many people on as many platforms as possible. I think there is a lot we can learn from each other, and a lot of duplicate effort that could be avoided, if we could just wrap our heads around that.
Slashdot Mirror
Hey Harv, apparently the moderators haven't had a chance to notice that this is one of your usual reposts to all *BSD threads. That's the only way this post could still be labeled 'interesting'.
Well, Harv, I don't see how your reply has anything to do with the statistics I posted. Nevertheless ...
> the FreeBSD project has only gotten more fractious and hateful in recent memory
You must not read the same mailing lists I do (most of the development ones). The last round of major flame wars died out around March/April and things have been much smoother since then.
Of course I don't expect this to influence you, as you're clearly only interested in repeating yourself without taking in new information.
> but the arrogance and condesendence of the *BSD community (and to be
> fair, it exists, to a much lesser extent, in the Linux community as well)
> has turned me off [...]
Well, I'll finesse the issue of trying to decide which community has more of this bad attitude by saying, why can't we just say this attitude isn't constructive, from anyone, period, and then just move beyond it?
My own personal opinion is that most people doing _real_ work on any of these projects are too busy to play the zealotry game. There's too much stuff to do. (In fact, I think I'll shut up on this thread now and go back to bufixes myself.)
That's right, msmith left because he didn't consider it fun anymore. Others have left too. Others stayed. Others joined. So what? that's the nature of volunteer projects. Why rehash this posting again and again?
It would be nice if you would attribute the original author of this email. IIRC it is from about a year ago. Some people left, some people stayed, life went on, as in any other volunteer project.
Boy, it looks like the parent post got under your skin much more than any other recent *BSD post I've seen. Instead of just posting your standard messages once per thread, you've posted each of them, what, 3 or 4 times in this thread already? I ask again, (as I asked earlier but not under this userid -- I have been too unmotivated to create a slashdot userid before), what is your goal in these repeated postings? Why not just avoid using the damned thing if you hate it? Why the repeated trolling with the exact same postings over and over again, in multiple threads, with differing subject lines?
> In the FreeBSD Ports Collection, there are many Ports marked as broken,
> and many more unmaintained and suffering from bit-rot.
Let me inject some facts into this discussion, as this is an area in which I've done a lot of research.
As of 12/30/2003, there are 10011 ports in the FreeBSD ports collection. Except for a very few cases where only a binary is available from the authors, the entire collection is meant to be buildable from source.
On 4-STABLE i386, 214 ports fail to build from source; on 5-CURRENT, 366 fail to build. The difference is primarily due to FreeBSD being an early adopter of gcc3.3 in 5-CURRENT. As these bugs are fixed, we try to encourage people to get them adopted upstream. This benefits the Linux as well as BSD communities.
239 ports are marked broken on 5-CURRENT; these are the 'hard-broken' cases. Right around 50% of those are compile problems and in some cases the code is indeed quite old. Some of the others that fail to build, but are *not* marked broken, are failing to fetch and part of that is due to the savannah compromise which is outside our control. That's temporary, we all hope. (I'm reluctant to see any port marked as broken unless it really won't compile/install -- not just having some transitory error.)
A slightly larger number, for which I don't have statistics, are marked broken on the 64-bit architectures, generally due to bad assumptions in their C code. Again, as these are fixed, changes make it back to the larger community. And, the number of broken ports on the 3 64-bit architectures has also come down dramatically.
The number of ports PRs has come down from a peak of 1500 during the 5.1-RELEASE freeze, to around 800 today. This is primarily due to a dozen new ports committers in the past 6 months who have been very active.
The number of officially unmaintained ports is 2616. Although this number may have decreased as well, clearly we still need more volunteers.
As for bit-rot, again, much of the code that is rotting lies outside the control of the FreeBSD ports team. Another project that is active is to try to identify, and prune, ports for which there is no hope. However, this requires getting a community consensus on what "no hope" is, and that takes time. The last time a pass was made scheduling ports for demolition, somewhere around 100 were proposed, and around 70 were fixed by someone or otherwise adopted in the 3 months afterwards. There are another 100 or so proposed for removal in the next pass; some of those have already been fixed as well. (I hope to create a framework for making this process more visible to end-users so there are as few "surprises" as humanly possible.)
One last point about bit-rot. For ports that fail to compile/install, the previous version of the binary ("package" in FreeBSD terminology) remains available for fetching until the new version works. Thus, most of the bit-rot only affects those who are installing from source.
The code that I've written that generates these statistics does so by mining the Problem Report database, the error logs from the bento build cluster (which iterates over each port on each architecture and each relevant OS release, continuously), the ports collection itself, and (to a limited degree) CVS meta-info. As far as I know no-one else has similar functionality. (Interested parties can contact me directly; since this machine is on the end of a cable modem line, I need to protect it from slashdotting). I hope to get this code up and running on a machine with greater bandwidth to make it more generally useful, soon.
To conclude, IMHO things that get fixed in the FreeBSD ports collection can have the effect of helping the greater community. I'd like to see the culture of zealotry start to disappear, and a greater degree of peer respect, especially in regards to making as many apps useful to as many people on as many platforms as possible. I think there is a lot we can learn from each other, and a lot of duplicate effort that could be avoided, if we could just wrap our heads around that.