Re: GRAB and APT (was: [[email protected]: Returned mail ...]
Greg Kurtzer <[email protected]>
| Newsgroups | gmane.linux.redhat.rpm.grab |
|---|---|
| Message-ID | <[email protected]> |
On Tue, Feb 19, 2002 at 10:28:42AM +0200, Michael Shigorin wrote: > > but I wanted to be easier to use (for the newbie), and not rely on specific > > Apt repositories (rather any FTP server). > Well, watching our distro (http://altlinux.com) switching to > APT/RPM from plain RPM, heavily ironed out for _right_ > dependencies repository is just vital for APT to work -- so it's > not exactly user's job to try and introduce APT, but rather > distributor's (who controls its package base). You have to remember that when I started GRAB, there was no Apt4RPM project that was publicly announced. And even now, there are still not that many Apt repositories on the net. I do agree that Apt is much more robust in solving dependencies, but GRAB is more of a tool used for finding packages and keeping your system up to date. GRAB will do its best to resolve deps when an installation bombs out, but that is just for the convenience of the user. > > Also, when I announced on the GRAB mailing list that the Apt > > project is finally moving to RPM, (with one exception), > Maybe "is being ported to RPM, too" instead of "is moving" :) I meant to say that... > > I created repositories, and rebuilt all of the Apt RPMS for > ^^^^^^^^^^^^ > > many OS's. I used it in place of GRAB everywhere I could. IT > Yes, and it's the most effort-taking part AFAIK. But this way > you become at least partially "co-vendor" :( I am not sure I follow... > > There is an anomaly of development of opensource code. It has > > to do with all of the focus being with the largest, most well > > known or _cool_ project out there. A lot of the other great > > ideas get lost because of lack of community support (i.e. > > Sendmail, Apache, WU-FTPD, etc...). These packages may not be > > the best, but they are historically the leaders and the > > community pays less attention to the "less well known" apps. I > Of these Sendmail is quite known for its idiosyncrasy, WU was > better when its authors nursed it (and more secure servers > instead of both these exist now) -- and I wouldn't say Apache is > out of focus or lesser-known software piece! My point was only some of the smaller projects that could replace these projects are not always given the same level of attention and contributions as the larger more well known projects. > > don't want to stop development on GRAB simply because there is > > a bigger, more well known app out there. GRAB may have similar > > functionality, but it also does other things that Apt has no > > future plans of doing. It is also easier to use for newbies. > No-no-no-no-no. Of course if significant effort has been put in > -and- community is around -and- it says they like it, why not? Simply saying "why not" is way too passive for my tastes. I did it because I saw a need. There was nothing else released in the community that worked the way I wanted. I was not asking for much, just a way to keep the systems I manage up to date, be able to find RPMS on the net, and streamline the installation/verification process. I released the code because I knew that if some members of the community also had a need for this type of an application they can also benefit, not to mention the role of community support in the form of code/idea contributions. > I personally don't think majority of distributions will switch to > APT promptly since (again) rep cleanup is *big* job to be done. It may be a big job, but because of the "lax" rules of spec'ing RPMS and RPM requires makes it a stressful job. This is an inherent disability in RPM IMHO. I do agree, that this is something the user should never see. > So easier, though probably not that strict and "done right" > implementations of the same basic principles are OK. Once again, I am not proclaiming GRAB to be a tool used for dependency resolution. I do a basic level (if the RPM was spec'ed out correctly by the packager) of dep resolution, but only because the info is there, so why not use it. GRAB was not designed for dep resolution. I have thought about adding a feature line that will do this, but why? Apt is there and does it VERY well. > Greg, don't take it up to heart -- it was never meant as > criticism but rather "but did you see that" :-) Don't worry about that. Of course I don't take it personally. > Say if I worked on some non-trivial task (realizing that doing it > right is many man-years and probably beyond my scope and > possibilities) _and_ you noticed the project and commented that > there's something better -- I'd only be thankful (I hope ;)! I am not sure I know what to say to this... Are you going to say the same thing to Philip Hazel (Author of Exim, Competitor to Sendmail)? Sendmail was around for many years before Philip started Exim (if I am not mistaken), and Sendmail was/is the defacto MTA on Unix. Philip I am sure saw something that he could either do better, or different. In any case, this was my earlier point. Simply because another project exists, does not mean there can not be competition. Because of the altruistic nature of the community, the best case scenario would be for both projects to work together, but that does not always happen (KDE/GNOME/WindowMaker/Enlightenment/FVWM...). > And I've almost stepped into that pitfall once. What I'd be > doing is far better done by gmp :-) Simply because a project has a large scope does not ever mean one should be discouraged from the undertaking. Simply because another project is currently doing it better, they can always use your ideas/code because of the open nature of the community. You may have been able to implement something better then the other project, or something they did not think of. They can borrow some of your code and vise versa, pretty soon both apps work extremely well. > > just believe that GRAB works as well, and fills certain voids > > that Apt leaves behind. BTW, I have approx 15,000 downloads of > > GRAB, (most from GRAB updating itself by users running it from > > CRON). > BTW what could be done easier if whole thing is already > apt-friendly? Could you spend a few words on that topic -- maybe > it would make sense to grab it into distro as well. What do you mean the whole thing? Do you mean the current repositories all over the net being Apt-friendly? Grab it into distro? I am not sure I understand... > > Good luck and strong health to you and your project! Thanks, and it is good to chat with you. I am looking forward to your reply! Greg BTW, what is your position/role with altlinux? -- Greg Kurtzer [email protected] http://www.runlevelzero.net LINUX,... The choice of a GNU generation! :wq