Re: [leaf-devel] uClibc-Bering and uClibc 0.9.27
Hi Gavin, Did you get the mail last night? If not, I can try again. It was very small in the end! I did get the mail, but it didn't contain a patch. I guess you didn't get my response to that (sent to your other account) ;-) Correct. We _have_ to move sooner or later, because we'll hit more and more packages that simply won't compile anymore with the old uClibc version (so far, we've managed to patch our way around that, but there'll be a point when that's impossible). Indeed. Related to that, it possible that effort will be wasted coding round problems that could have been resolved in moving to the later version It's just a question of working out where the breakeven point of effort is. Right. And often it is easier (or rather, the risks are easier to control) to have to work on one specific package so it will compile against an old uClibc, than switching to a new uClibc and having to make sure everything still compiles and works (the latter is obviously more difficult, since that requires a lot of testing - and some of the packages that were built upon request by users aren't even used by anybody in our group, so it's extremely hard to test those). And then, the newer versions also offer additional features (like the NPTL support that's currently being worked on). The new feature around the corner! I also think a new libm is expected as well. However, will we be able to use NPTL without a 2.6 kernel? Well, RedHat made it work with RedHat 9 and Fedora Core 1, so it is possible - but I don't know if NPTL support is there in the vanilla kernel. Plus will it actually be that useful for what these systems do? Good question - but I've given up thinking about that a long time ago - different users come up with different usages for LEAF (just look at the list of packages we provide - there are quite a few that I'd never even let come close to my router ;-)). But since not every LEAF installation is a router, it's ok with me to at least test if those new features can make it into Bering uClibc (within reason - we're not willing to waste too much space just to add a new feature hardly anybody will use). Martin -- You think that's tough? Try herding cats! --- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
RE: [leaf-devel] uClibc-Bering and uClibc 0.9.27
Hi Martin, Did you get the mail last night? If not, I can try again. It was very small in the end! >Charles already answered that one - but if you want to find out even >more than what he wrote, have a look at >http://en.wikipedia.org/wiki/Application_binary_interface Useful site that. Thanks! >Correct. We _have_ to move sooner or later, because we'll hit more and >more packages that simply won't compile anymore with the old uClibc >version (so far, we've managed to patch our way around that, but >there'll be a point when that's impossible). Indeed. Related to that, it possible that effort will be wasted coding round problems that could have been resolved in moving to the later version It's just a question of working out where the breakeven point of effort is. >And then, the newer versions also offer additional features (like the >NPTL support that's currently being worked on). The new feature around the corner! I also think a new libm is expected as well. However, will we be able to use NPTL without a 2.6 kernel? Plus will it actually be that useful for what these systems do? >You mean, you used binaries compiled against uClibc 0.9.20 with uClibc >0.9.27? If that worked, you've just been very lucky (since that usually >ends in a segfault rather quickly). Knowing what I do now, I tend to agree with you :-) Gavin PS I have no control below here!! Sorry ** Visit the Virgin Atlantic website for all the latest news and fantastic offers - http://www.virgin.com/atlantic This e-mail (and any attachments) may contain privileged and/or confidential information. If you are not the intended recipient please do not disclose, copy, distribute, disseminate or take any action in reliance on it. If you have received this message in error please reply and tell us and then delete all copies on your system. Any opinion on or advice or information contained in this email is not necessarily that of the owners or officers of this company. Should you wish to communicate with us by e-mail, we cannot guarantee the security of any data outside our own computer system. ** --- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] uClibc-Bering and uClibc 0.9.27
Hi Gavin, [EMAIL PROTECTED] wrote: Hi Martin, Yes, it does look like they have broken it again. BTW, do you know what ABI means? I've had a hunt, but it appears to be an acronym that you either know the meaning of or you don't... and I don't! Charles already answered that one - but if you want to find out even more than what he wrote, have a look at http://en.wikipedia.org/wiki/Application_binary_interface > Fundamentally, from the IT techie perspective, it normally boils down to "what does the new version give us that the old version doesn't" or "should we move to the new version because we have sat on the old one for such a long time". The first is probably a valid question here, but I can't answer that (doing this more for fun than anything else!). The second is less relevant, unless we have support issues, which again might not be the problem here. Correct. We _have_ to move sooner or later, because we'll hit more and more packages that simply won't compile anymore with the old uClibc version (so far, we've managed to patch our way around that, but there'll be a point when that's impossible). And then, the newer versions also offer additional features (like the NPTL support that's currently being worked on). For what it is worth (and I have only tested limitedly) the old binaries for other things appeared to work... You mean, you used binaries compiled against uClibc 0.9.20 with uClibc 0.9.27? If that worked, you've just been very lucky (since that usually ends in a segfault rather quickly). Martin --- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
Re: [leaf-devel] uClibc-Bering and uClibc 0.9.27
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 [EMAIL PROTECTED] wrote: | Hi Martin, | | Yes, it does look like they have broken it again. BTW, do you know what | ABI means? I've had a hunt, but it appears to be an acronym that you | either know the meaning of or you don't... and I don't! I believe it's Application Binary Interface, but whatever it stands for, it's referring to the calling convention required to use the code. Note that there can be several *DIFFERENT* ways to interface (pass parameters & return results) to code (ie: register(s), stack(s), pointer/value passing, etc), and everyone on both 'sides' has to agree on the conventions. In the context of a uClibc, the low-level calling conventions between routines are standardized (by gcc), and ABI is most likely referring to the interface visible to the application (ie: how many and what kind of parameters are required for each function, and what result type (if any) is returned...the stuff you find in the .h header files), which would require a recompile if it changed. - -- Charles Steinkuehler [EMAIL PROTECTED] -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.0 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFCndc6LywbqEHdNFwRAkMUAKCfV0dMjcs3OcxG7tg1gC4Bl1/BoACfWHXP hgV1W5uabMJ0WgeXGpdz9J0= =IMik -END PGP SIGNATURE- --- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
RE: [leaf-devel] uClibc-Bering and uClibc 0.9.27
Hi Martin, Yes, it does look like they have broken it again. BTW, do you know what ABI means? I've had a hunt, but it appears to be an acronym that you either know the meaning of or you don't... and I don't! I guess it's the normal problem of do I upgrade to the latest thing, when I know that there is another (even greater) thing coming straight after it. I have that all the time here at work! Fundamentally, from the IT techie perspective, it normally boils down to "what does the new version give us that the old version doesn't" or "should we move to the new version because we have sat on the old one for such a long time". The first is probably a valid question here, but I can't answer that (doing this more for fun than anything else!). The second is less relevant, unless we have support issues, which again might not be the problem here. For what it is worth (and I have only tested limitedly) the old binaries for other things appeared to work... However, there are better minds than mine that need to check this! Gavin PS I have no control below here! Sorry ** Visit the Virgin Atlantic website for all the latest news and fantastic offers - http://www.virgin.com/atlantic This e-mail (and any attachments) may contain privileged and/or confidential information. If you are not the intended recipient please do not disclose, copy, distribute, disseminate or take any action in reliance on it. If you have received this message in error please reply and tell us and then delete all copies on your system. Any opinion on or advice or information contained in this email is not necessarily that of the owners or officers of this company. Should you wish to communicate with us by e-mail, we cannot guarantee the security of any data outside our own computer system. ** --- This SF.Net email is sponsored by Yahoo. Introducing Yahoo! Search Developer Network - Create apps using Yahoo! Search APIs Find out how you can build Yahoo! directly into your own Applications - visit http://developer.yahoo.net/?fr=offad-ysdn-ostg-q22005 ___ leaf-devel mailing list leaf-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/leaf-devel
