Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
Linux Software

Marcelo Tosatti on UnitedLinux (And More) 134

Posted by CmdrTaco from the stuff-to-read dept.
PDAJames writes "There's a new interview with Marcelo Tosatti online. He talks about what it's like to be kernel maintainer after nine months on the job and identifies the biggest problems with the kernel now. He also defends UnitedLinux against Mandrake's criticism. He would, since he works for Conectiva..."
This discussion has been archived. No new comments can be posted.

Marcelo Tosatti on UnitedLinux (And More) More

Marcelo Tosatti on UnitedLinux (And More)

Comments Filter:
  • It's not like working on 2.5 (the development kernel).
    Yeah, that kind of thing is really sexy.

    Marcelo is a kernelophile! He should be locked up for having feelings like towards kernels! ;)
    • it's not illegal to say you have feelings twords kernels, though in Penn. your ISP might be required to report you saying such to the state authorities. let's just hope the traffic between /. and Marcelo doesn't have to hop through pittsburg ;).
    • Marcelo is a kernelophile!
      Look
      here [linux.org] and read about Linus' thoughts about how Tux should look (this is from the early days before Linux had a logo).

      After you've read the whole article, it should be quite obvious that Linus has a "thing" for penquins. If you've read up on Linus at all you would know that it probably borders on a fetish.

      Now re-read that last paragraph and ask yourself- what exactly was Linus doing that had him being chased by a speeding penguin..

      I think the marrage was just a publicity stunt, much like when Troy McLore married Thelma ..


    • It's just a cool phrase. My boss at work uses it all of the time. Whenever I jazz something up because it is going outside of our production to someone else who has to look at it, and he likes it, he will say "Man, now that is sexy. " Or when I am spending too much time formatting when I should be just digging it getting the problems solved, he will say "Hey, there's no need to make it all sexy, just get the problem solved and then worry about all of that other crap later." So, it's just a cool phrase, and I think the maintainer is a pretty cool guy too."

  • Unfounded criticism (Score:2, Insightful)

    by RocketJeff ( 46275 )
    "He also defends UnitedLinux against Mandrake's criticism. He would, since he works for Conectiva..."

    Yes, I know it from the submitter (not an editor), but it's still a dumb comment.

    Just because you're an employee doesn't mean that you can't have opinions and express them. If anything, I'd think an employee of a company dealing with Open Source software would be more likely to express dissatisfaction when warranted. I suppose PDAJames (the submitter) only says nice things about the company he works for?

    • Now a days Tossati is so influent not only in Brazil, but also worldwidely, that he doesn't need to agree with Conectiva. He agrees only if he wants to.

      Today Tossati can open his own linux company and be his own boss for the rest of his life. If he's saying that UnitedLinux is cool then it is cool for him (for him).

      Besides this, what the heck! As stated by GigsVT here [slashdot.org] Tossati develops the kernel, not applications, nor distros.

  • Heh (Score:5, Insightful)

    by GigsVT ( 208848 ) on Thursday July 11, 2002 @09:06AM (#3863944) Journal
    There are some simple problems that non-technical users might have with Linux, such as installing new software, or finding certain files that they need in the system. What's the solution to that?

    Why do people keep insisting on asking kernel maintainers questions that only related to user level apps? What's the point? That's like asking the Gaim team about the newest block device code.

    • Re:Heh (Score:2, Insightful)

      by freuddot ( 162409 )
      Why do people keep insisting on asking kernel maintainers questions that only related to user level apps?

      If you accept these two facts (er. assumptions):

      a) The kernel is not the UI and as nothing to do with usability
      b) The main thing keeping Linux out of the desktop market is that Joe Public is unable to USE it.

      Then, you must accept that the main reason why Linux doesn't work on the desktop market has nothing to do with Linux itself, but rather is an usability problem.

      If you don't accept that, then you must assume that kernel maintainers have something to do with Linux success, and by extension with the UI / UserLand.

      To answer your question, how about people on /. are not ready to accept this ?

      • > a) The kernel is not the UI and as nothing to do
        > with usability

        You're mostly right, however there are some overlapping areas.

        For example (cheap shot, I know :) take CD burners.

        Why the fsck should I need scsi-ide emulation to have my 100% IDE burner act as if it were an SCSI one ??? (and hang the SCSI stuff once every now and then, btw)

        Answer: (as far as I understand it, IANAKH)
        because at the time it was easier to "route" the IDE
        burners through the SCSI API instead of moving the common code to the upper level.

        But try to explain to Joe User why he should
        add hdc=ide-scsi to his lilo.conf ! He'll find that dumb, and I can't blame him.

        That's just an example, but it *does* mean that kernel and "user experience"(tm) are not separate things.

        • *sigh*, once again ... we have people living in the past.

          Grab a modern linux distro, install ... wow, look, burner recognized, icon on desktop ... Click, burn.

          User doesn't know/care about any of what you just mentioned.
          • heh, lol :)

            It's true I haven't tried consumer-level distros in a while. That's good if they get that one right.

            However, kindly allow me to modify my point in:

            When the sh*t hits the fan and Joe User tries to understand why his drive won't respond, won't he feel weird getting those SCSI errors ?

            (Besides, think of the poor Mandrake/Suse/RHAT dudes that do the behind-the-scenes configuration stuff ! I think they'd like a more consistent behavior from the kernel :)
            • When the sh*t hits the fan and Joe User tries to understand why his drive won't respond, won't he feel weird getting those SCSI errors ?
              Right ... we should try to emulate those clear and concise Windows error messages like "General Protection Fault" or "Error Reading Drive". Better yet, go down Apple's road and just print a sad looking penguin, or a bomb, with a hexadecimal error number.
              • Right ... we should try to emulate those clear and concise Windows error messages like "General Protection Fault" or "Error Reading Drive". Better yet, go down Apple's road and just print a sad looking penguin, or a bomb, with a hexadecimal error number.

                You're saying the only 2 possibilities are the way Linux is now or the way Windows is? Because the author of the comment above was only saying it needs to be better. This kind of knee-jerk reaction shows how a lot of Linux advocates only think in terms of Linux vs. Windows; I'm not saying comparisons shouldn't be made to Windows, when appropriate, just that there some things that are just not good enough in ANY desktop system today and suggestions about how such things could be improved in Linux stand on their own.

                -chris
                • Thats not how I interpreted it. I read him as saying "People would rather use Win than Lin, 'cos the latter gives you cryptic error messages." I was observing that this was true of both. Giving meaningful ones is a hard problem, and would require hardware drivers to become self-diagnosing expert systems. Not gonna happen, I'm afraid.

        • Re:Heh (Score:2, Insightful)

          by mark_lybarger ( 199098 )
          guess what... windows cd-burning software also uses scsi emulation. that's what the atapi is all about on those cd-rom drives. it give software developers a common interface to program to.

          which "upper level" are you referring to? to my understanding the (scsi) emulation layer is an upper level. this makes all the code common. as others have mentioned, most big distros will recognize your cdrw and load the scsi modules on boot. launch xcdroast and your in business.
          • > which "upper level" are you referring to?

            I seem to recall seeing somewhere that there were plans to move some of the SCSI code (incl. that which is used by ide-scsi) that doesn't specifically belong to SCSI to a higher, more abstract level. This way, devices that need SCSI-*like* functionality don't feel the drag of the whole SCSI interface.

            But I'm not a specialist, so check google for that. (I just did, but I've got waaaay too many results :)

            • i guess i'm too lazy to search google for "remove code scsi higher level". ok. ok. ok.

              the original post was regarding the complexity of using linux scsi drivers to access ATAPI CDROM drives, specifically CDRW drives. i don't see the relavence of moving any code, or the "drag of the whole SCSI interface".

              at any rate, you seem to be suggesting an ide-super-duper-scsi interface. how would using this interface make setting up the device for the user any easier? they're still not using an IDE device, and the applications are all written to take advantage of the SCSI interface. maybe if the ide-super-duper-scsi interface would talk to both ide and scsi devices, and applications would use the new interface, we'd be all set.

        • Re:Heh (Score:4, Interesting)

          by GoRK ( 10018 ) on Thursday July 11, 2002 @11:43AM (#3864969) Homepage Journal
          Well you are wrong about one thing: "100% IDE"

          Burners are ATAPI devices - Calling them "IDE" is sort of like saying "Ethernet" when you mean "http" or something else like that. ATAPI was invented to bring SCSI devices into the consumer market with minimum cost to hardware manufacturers (ie no change of command api's in the devices firmware). As far as the command set goes, they are actually almost 100% SCSI. Ditto with ATAPI zip drives and tape drives. Heck, even the parallel port zip drives are SCSI (with a built in ppa SCSI controller). USB mass storage uses the SCSI command set also -- and I think Firewire does too (I may be wrong about firewire). ide-scsi is *not* a SCSI emulator. It is a transport, the same as the USB mass storage transport.

          IMO, the mistake was to write a seperate driver for ATAPI in the beginning that did NOT tie to the SCSI system - we might have a mature ATAPI/SCSI interface by now that did not have the problems that we see today.

          And why do you mention only CD burners? I use ide-scsi for regular CDROMS, zip drives, and tape drives. The application support is much better (ie i can do cd to cd copies, use all backup software, etc)

          ~GoRK

        • Regarding the IDE-SCSI necessity for CD writers
          it should be noted that CD writers are NOT plain
          IDE. This is why on windows you need ASPI from
          adaptec (now standard).

          There should be some easier solution, but IDE-SCSI
          is necessary.

          Petros

      • Re:Heh (Score:2, Insightful)

        by CarrionBird ( 589738 )
        How about hardware drivers?

        For example: To a convert who's used to one WDM driver working (to a degree) for 98, me, 2k, xp, WhatEver, It doesn't make much sense that his video driver must be recomplied for every version of every distro. Methinks people from the MacOs side would think the same thing.

        It may not be a big deal for a developer type person, but someone who's just want's to try Linux out & isn't a programmer will have a hard time getting his 3d card going. So, a lot of guys walk away thinkng "Well Linux can only do 3d if you have X card with Y distro and stand on one foot, so I'll just stick with Lameos 99 which at least will support XYZ card.".

        Even if you don't care about the unwashed masses, we're all affected by the lack of vendor support that results.

  • Linux Magazine interview (Score:3, Informative)

    by larry bagina ( 561269 ) on Thursday July 11, 2002 @09:08AM (#3863958) Journal
    devloperWorks has this [ibm.com] interview with Marcello from March. Ignore the Java exceptions :)

  • Can't we all just get along (Score:3, Interesting)

    by ShwAsasin ( 120187 ) on Thursday July 11, 2002 @09:10AM (#3863970) Journal
    Although I am Mandrake/SuSe/RedHat user, I still find each distro great in their own individual way. I think United Linux offers an interesting solution to the bigwigs (Red Hat, and Mandrake on a lessser scale) but it will be tough for them especially when RH/Drake have both the server / desktop markets in their palms.

    But we all know that things can change in a 90 degree turn on a day, i.e. Sega Saturn vs. PS, Genesis vs. SNES, OS9 vs. OSX, etc...

    Only time will tell.
    • if only you had mentioned debian/gentoo distros. the source is stong in that one i can tell, but there is still much fear from the masses in building their own box...

      a box is a box. when it all boils down to it, do the applications work? do they work together? does it get the job done? ...
  • I think that he has made some good points about the value of UnitedLinux. It's nice having all the choice when selecting your distro, but for software companies writing for linux systems has to make sure that they can run on all distros is certainly an expensive undertaking.

    And I think that pushing embedded linux is also an excellent point. It is a marked that linux can take over (unlike the current desktop).

    I was disapointed he didn't talk to much about linux on the desktop, and what his views are about where/how to proceed. Anyone know of other interviews of his where he mentioned them?

  • Wise words from a young man (Score:5, Insightful)

    by red_dragon ( 1761 ) on Thursday July 11, 2002 @09:18AM (#3864043) Homepage

    This bit really stood up, IMO:

    It's just technology, so why do these clashes come up, do you think?
    Because people are there. Egos show up.

    And it could not be explained any more simply than that.

    • Re:Wise words from a young man (Score:1, Funny)

      by Anonymous Coward
      Not so wise words, paraphrased:

      I've recently learned that saying, "I think you're wrong." is much more socially adept than saying, "You stupid piece of shit."

  • memory corruption (Score:1, Funny)

    by Anonymous Coward
    Tosatti: Somewhat. My priorities are fixing some corruption bugs -- we're not sure if they're a hardware problem or not. We're not sure where the thing is, or whether it's present on all kernels, but it's memory corruption, which is a big problem because memory corruption may cause data corruption and crashes.
    Oh dear! This might explain the frequent kernel errors. Hopefully, they'll implement "protected memory" someday!

  • Marcelo (Score:4, Insightful)

    by MrHat ( 102062 ) on Thursday July 11, 2002 @09:22AM (#3864067)
    I just try to be as technical as possible. It is very important to me... People do not use the right words some of the time. Starting a fight is really easy. But instead of saying, for example, "You are full of shit," you can just say, "I don't think you are really right." You have to use the right words.

    Slashdot could learn a lot from this guy.

    • Slashdot could learn a lot from this guy.

      No we can't, you dumbass! You're completely full of shit!

      ;-)

    • Linus (Score:5, Funny)

      by bgarcia ( 33222 ) on Thursday July 11, 2002 @11:10AM (#3864762) Homepage Journal
      But instead of saying, for example, "You are full of shit," you can just say, "I don't think you are really right." You have to use the right words.
      Slashdot could learn a lot from this guy.
      Hell, Linus [google.com] Torvalds [google.com] could learn a lot from this guy.
      • Hell, Linus [google.com] Torvalds could learn a lot from this guy.

        Heh, very true, especially as Linus was at least partly wrong in the first of those tirades: buffers couldn't handle buffers bigger than page size then, and still can't, even today.

        The trouble is, the audience tends to love this mudslinging, and that just encourages the behaviour.
    • I believe they have.

  • Can't go to the U.S.? (Score:3, Interesting)

    by Ami Ganguli ( 921 ) on Thursday July 11, 2002 @09:32AM (#3864113) Homepage

    He said he got turned back at the border once for having a B1 visa when he needed a B2, and now he can't go back. What's the deal here? Is he banned for life?

    • Re:Can't go to the U.S.? (Score:3, Informative)

      by Anonymous Coward
      To be allowed to enter the US after being rejected on an incorrect visa, he needs to go through some long bureaucratic process which may or may not result in the ban being lifted. He probably figures that it isn't worth doing until he has a really good reason to go to the US.
      • He probably figures that it isn't worth doing until he has a really good reason to go to the US.

        Well, we provide no good reason, unless you like to have Mr. Gates and Mickey rule everything you do. I'd call it a gift.
      • To be allowed to enter the US after being rejected on an incorrect visa, he needs to go through some long bureaucratic process which may or may not result in the ban being lifted.

        That's exactly true, starting with the fact that the u.s. immigration service won't even talk to anybody who'd not a u.s. citizen. You Americans have no idea what a bad impression your representatives are making at the border, if you did you'd demand changes.
    • It isn't a big deal to have it fixed, but you do have to go to your American consulate in your region and stand in a couple of lines. Act dumb. Be nice. Be confident. Be contrite.

      He should take care of it now as it may take a few months for the paperwork to get done (or a few hours, depending...) and that may obstruct him if he wishes to go to the US for some odd reason.

    • Re:Can't go to the U.S.? (Score:4, Insightful)

      by 4of12 ( 97621 ) on Thursday July 11, 2002 @11:15AM (#3864791) Homepage Journal

      He said he got turned back at the border once for having a B1 visa when he needed a B2, and now he can't go back. What's the deal here?

      Uhm, you haven't been following the earlier posts which clearly explained how Marcelo was much wiser for his years in terms of dealing with people and personalities as well as being a technically adept kernel hacker and having the ability to think about a huge number of diverse kernel patches being applied to a production system used by millions worldwide.

      Shoot, if the United States were to start letting people like Marcelo into the country there's no telling what kind of mess we'd be in!

      Now, if he were slyly willing to work off the books in a meat packing plant or a demolition site for forty dollars a day he might have a better chance...

  • In the interview, Marcelo says that because he mistakenly entered the U.S. on business, using a tourist visa, he was turned away and now cannot reenter.

    Is this a permanent condition? And is there any way to appeal this? Contact INS or our local representatives?

    I understand the government's paranoia post-911, but this is throwing the baby out with the bathwater.

  • Hey I am not a programmer but this doesn't seem to be the most efficient way of doing things? Has this efficiency question ever been raised??? enlighten me linux hackers

  • akpm (Score:3, Interesting)

    by neroz ( 449747 ) <neroz@iinet.ne t . au> on Thursday July 11, 2002 @10:03AM (#3864301) Homepage
    No offense to Marcelo, 2.4.18 was a great release.. but I honestly think Andrew Morton would have been a better choice for maintainer. The delay between 2.4.18 and 2.4.19 has been several months long - a few months is acceptable because of the IDE changes, but this is just a bit much. The time between pre-releases has been too long recently as well - the longer between each pre, the more changes that are made between them == less testing of the changes seperately.

    (Yes, I know all the arguments about how it is a "stable" series and all - but some people want new features, and 2.5 is not an option, thanks to IDE being 0wned by Martin).

    This is _NOT_ a troll.

    • Just one question...

      • Do you prefer to have a dead kernel release that always crashes your filesystem?

    • Re:akpm (Score:3, Interesting)

      by guybarr ( 447727 )
      I know all the arguments about how it is a "stable" series and all
      but you want to discard stability of a STABLE branch for features - not a great deal for users who wish to actually use the system.

      - but some people want new features, and 2.5 is not an option, thanks to IDE being owned by Martin
      have you considered that IDE instabilities are in 2.5 because this is where new features and engineering reforms should be done, regardless of who is the maintainer ?

      my belief: there should be a split to 3 branches:
      1) long-term dev-branch: this branch is where large-scale (>months) changes are made and tested.
      2) medium and short-term dev-branch: this branch is what you're looking for, where features and fixes which take months or less are made and tested, really tested, not like the 2.4 VM fiascos, before becoming stable.
      3) stable branch: this branch changes once a year or less, except for very minor bug/security fixes.

      changes will NOT be imported directly from 1->3, only 1->2 , 2->1 and 2->3 .

      this will give you both stability for those who need it, long-term dev and testing environemnt for the systemic reformers, and short-term devellopment for the medium- to small-scale changes.

      and before anyone of you jumps and shouts "but this is exactly the debian model of stable, testing, unstable branches" I say that both branch 1 AND 2 are dev. AND testing branches, the criterion sugested is the changes scale, not the targeted audience (though ususally decisions may be similar between models).
      • I agree with everything you said, other than the IDE comment. What should have happened with IDE, is [as was stated by Jens, who recently forward-ported 2.4's IDE to 2.5], there should have been both the 2.4 IDE and the 2.5 IDE in the tree, with a kernel option to pick which one to use. That way people can use the kernel for non-IDE development and not have to worry about their harddrives melting. :-) I think now they will switch to that, otherwise Jens will have to maintain his port himself..

        And I'll say it again, a longer release cycle does not mean the end result will be good. Because there is a large gap between releases right now, a lot of changes are going into each pre/rc - which means more chances for breakages. 2.4.18 was a really good release, and 2.4.19 should have built upon that. [IMO, the new IDE should have gone into 2.4.20-pre*].
      • my belief: there should be a split to 3 branches: 1) long-term dev-branch: this branch is where large-scale (>months) changes are made and tested. 2) medium and short-term dev-branch: this branch is what you're looking for, where features and fixes which take months or less are made and tested, really tested, not like the 2.4 VM fiascos, before becoming stable. 3) stable branch: this branch changes once a year or less, except for very minor bug/security fixes. changes will NOT be imported directly from 1->3, only 1->2 , 2->1 and 2->3 .

        You do realize, right, that this is exactly the FreeBSD release method? (3) is STABLE, (2) is RELEASE, and (1) is CURRENT.
        • You do realize, right, that this is exactly the FreeBSD release method?
          wrong, I actually don't know the BSD dev-model.

          (3) is STABLE, (2) is RELEASE, and (1) is CURRENT.
          but if I understand correctly the meaning of RELEASE label, it is for testing purposes only, while I suggest 2 branches for both dev and QA, only the scales differ.

          then when you approach a "freeze" (updating 1->2 or 2->3 ) you will need a short-lived release-candidate branch for each.

  • Attitute Problem? (Score:3, Insightful)

    by blancolioni ( 147353 ) on Thursday July 11, 2002 @10:10AM (#3864345) Homepage
    He also defends UnitedLinux against Mandrake's criticism. He would, since he works for Conectiva...

    Marcelo was polite and reasonable throughout the interview, and gave clear reasons for his position. What's with the bitchy editorialising?
    • Attitute? I suck at this.

      See, I'm even disagreeing with myself! Point proven.
  • There are a lot of potentially clashing personalities in the Linux world. How do you deal with them?
    I just try to be as technical as possible. It is very important to me... People do not use the right words some of the time. Starting a fight is really easy. But instead of saying, for example, "You are full of shit," you can just say, "I don't think you are really right." You have to use the right words.
    He forgot to mention however that this kind of atitude does not apply on the DJB lists.(qmail@, djbdns@, etc) hehe

    -davidu

Slashdot Top Deals

Solutions are obvious if one only has the optical power to observe them over the horizon. -- K.A. Arsdall

Close