Results 1 to 7 of 7

Thread: klik:// pseudo protocol, KDE integration, MIME Types

  1. #1
    Junior Member
    Join Date
    Dec 2004
    Posts
    6

    klik:// pseudo protocol, KDE integration, MIME Types

    Despite unclear security measures with it (this system makes it probably as easy for Linux newbies to get a trojan as it is already for Windows noobs), I pretty much like the idea of one-click installations and fully stand-alone and encapsulated program packages.

    I'd however like to point out (nitpicking of course) that your use of that "klik://" pseudo protocol is very much against all standards. I suspect it is because your primary or only target are KDE users, and it is to hook into there much easier using new protocol-prefixes. To be standards compliant (WWW, XML, XHTML especially) you should however only use "x-klik://" to explicitely mark it as experimental and non-standard, so it won't yield problems.

    A better alternative was to abandon that redundand prefix, even if it looks so nice at a first glance. MIME Types are more appropriate for that task anyhow, and would allow to hook the klik system into nearly any Web browser and outside of the KDE landscape (which pretty much seems to be the only system suffering from too many of those pseudo protocol prefixes).

    For example "application/x-klik" would be an easy way to go. Additionally that klik server could send it with a header of "Content-Type: application/x-klik; depends=glibc2.3,mplayer; arch=i686" or so. I think this is superiour to any pseudo protocol prefix klik:// which in fact is only an alias to HTTP. And I don't think a mime type is any harder to chain into KDE than such an URI prefix.

    Else, a great idea.

  2. #2
    Senior Member registered user
    Join Date
    Aug 2003
    Location
    Dublin, Ireland
    Posts
    164

    Re: klik:// pseudo protocol, KDE integration, MIME Types

    Quote Originally Posted by tmp000
    Despite unclear security measures with it (this system makes it probably as easy for Linux newbies to get a trojan as it is already for Windows noobs), I pretty much like the idea of one-click installations and fully stand-alone and encapsulated program packages.

    I'd however like to point out (nitpicking of course) that your use of that "klik://" pseudo protocol is very much against all standards. I suspect it is because your primary or only target are KDE users, and it is to hook into there much easier using new protocol-prefixes. To be standards compliant (WWW, XML, XHTML especially) you should however only use "x-klik://" to explicitely mark it as experimental and non-standard, so it won't yield problems.

    A better alternative was to abandon that redundand prefix, even if it looks so nice at a first glance. MIME Types are more appropriate for that task anyhow, and would allow to hook the klik system into nearly any Web browser and outside of the KDE landscape (which pretty much seems to be the only system suffering from too many of those pseudo protocol prefixes).

    For example "application/x-klik" would be an easy way to go. Additionally that klik server could send it with a header of "Content-Type: application/x-klik; depends=glibc2.3,mplayer; arch=i686" or so. I think this is superiour to any pseudo protocol prefix klik:// which in fact is only an alias to HTTP. And I don't think a mime type is any harder to chain into KDE than such an URI prefix.

    Else, a great idea.
    I'm afraid I must post out that in fact the reason for the pseudo protocol is because it is not simply an alias for http but in fact signifies that the hostname/programname should be given to the klik program as an argument, the klik program then gets and runs the recipe. Perhaps instead a mime-handler could be added for recipes, but it doesn't leave the user with as simple a scheme to remember and it doesn't mean you have a command line klik app that can be used identically. Also it is worth noting klik support is implemented/imminent for elinks and firefox.

    Not argue about whether or not it is the correct way to do it (standards) just want to draw your attention to a few things.

  3. #3
    Senior Member registered user
    Join Date
    Feb 2003
    Location
    Germany
    Posts
    1,159

    Re: klik:// pseudo protocol, KDE integration, MIME Types

    Hi tmp000,

    thanks for your input which we appreciate.

    Quote Originally Posted by tmp000
    this system makes it probably as easy for Linux newbies to get a trojan as it is already for Windows noobs
    It doesn't, exactly for the reason that we do not use MIME Types. If we were using MIME Types, then anyone could set up a server to send the neccessary MIME headers, and in turn anyone's recipes would be run by the klik client. But since we are not using MIME Types, the klik client does not run anything except recipes from the official klik server. Please note that klik URLs do not contain a hostname, so they are not only easier to remember, but also it's impossible to trick the user with modified klik links. In fact, it's just a matter of trusting the klik server or of not doing so (as with any server you are downloading from).

    In summary, "it's not a bug, it's a feature"

    Greetings,
    probono

  4. #4
    Senior Member registered user
    Join Date
    Feb 2004
    Posts
    949

    Re: klik:// pseudo protocol, KDE integration, MIME Types

    Quote Originally Posted by probono
    Hi tmp000,

    thanks for your input which we appreciate.

    Quote Originally Posted by tmp000
    this system makes it probably as easy for Linux newbies to get a trojan as it is already for Windows noobs
    It doesn't, exactly for the reason that we do not use MIME Types. If we were using MIME Types, then anyone could set up a server to send the neccessary MIME headers, and in turn anyone's recipes would be run by the klik client. But since we are not using MIME Types, the klik client does not run anything except recipes from the official klik server. Please note that klik URLs do not contain a hostname, so they are not only easier to remember, but also it's impossible to trick the user with modified klik links. In fact, it's just a matter of trusting the klik server or of not doing so (as with any server you are downloading from).

    In summary, "it's not a bug, it's a feature"

    Greetings,
    probono
    Very well thought out, the only thing that could screw this up is users downloading a "unoffical" version.

  5. #5
    Junior Member
    Join Date
    Dec 2004
    Posts
    6

    more open use

    I see how it makes the URLs nicer and the whole system unbreakable, if any hostname is omitted and there is only one valid server / source for all klik recipies. But then, I cannot let the argument count, that it should be "easy to type and rembember" for users, if the original idea was to make (mouse) one-click installations (no keyboard involved here ;-)

    Also it would be sad to have such a thing closed to one project site. You could still open it up, and mandate security with a domain withelist - like Mozilla/Firefox do, only that you should make it harder and show more warnings for adding new recipe sources. I believe there is more value in this than just serving a few Knoppix-derived distributions.

    If you decide to keep the current (hostname-free) links, then at least get rid of the double slash // - as the klik-links then are clearly URNs and not Urls. "urn:klik:PACKAGE" was the best choice then (with the "x-klik" variant to become a standards addicted).

    Best,
    mario

  6. #6
    Senior Member registered user
    Join Date
    May 2003
    Location
    Utrecht, The Netherlands
    Posts
    298

    Re: more open use

    Quote Originally Posted by tmp000
    I see how it makes the URLs nicer and the whole system unbreakable, if any hostname is omitted and there is only one valid server / source for all klik recipies. But then, I cannot let the argument count, that it should be "easy to type and rembember" for users, if the original idea was to make (mouse) one-click installations (no keyboard involved here

    Also it would be sad to have such a thing closed to one project site. You could still open it up, and mandate security with a domain withelist - like Mozilla/Firefox do, only that you should make it harder and show more warnings for adding new recipe sources. I believe there is more value in this than just serving a few Knoppix-derived distributions.

    If you decide to keep the current (hostname-free) links, then at least get rid of the double slash // - as the klik-links then are clearly URNs and not Urls. "urn:klik:PACKAGE" was the best choice then (with the "x-klik" variant to become a standards addicted).

    Best,
    mario
    I guess klik:// is just a KIO slave. the fact gnome and other DE's don't support them (altough support is probably going into the kernel, KIO-fuse) is their fault, imho. Its a great piece of technology, its free software, so its totally up to other environments/filebrowsers/whatever to use it or not.

    klik:// is a protocol, not a url. just als locate://, smb://, http://, ftp://, apt://, settings://, tar:// etcetera.

  7. #7
    Senior Member registered user
    Join Date
    Feb 2003
    Location
    Germany
    Posts
    1,159

    Re: more open use

    Quote Originally Posted by tmp000
    Also it would be sad to have such a thing closed to one project site. You could still open it up, and mandate security with a domain withelist - like Mozilla/Firefox do
    We are going into that direction with the "Contributed kliks" page:
    http://klik.atekon.de/contrib.php

    Greetings,
    probono

Similar Threads

  1. Klik is not a registered protocol
    By emlee5 in forum Klik
    Replies: 6
    Last Post: 07-04-2006, 08:55 PM
  2. Install types
    By rniles in forum Hdd Install / Debian / Apt
    Replies: 2
    Last Post: 12-24-2004, 05:07 PM
  3. Replies: 2
    Last Post: 12-08-2004, 11:23 AM
  4. Device types on KDE: Too few?
    By fcc in forum Customising & Remastering
    Replies: 1
    Last Post: 12-24-2003, 03:40 PM
  5. Replies: 16
    Last Post: 05-21-2003, 10:38 PM

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  


LSI 9305-16i SATA SAS 12Gbs RAID Controller PCIe 3.0 x8 IT-Mode 4* 8643 SATA picture

LSI 9305-16i SATA SAS 12Gbs RAID Controller PCIe 3.0 x8 IT-Mode 4* 8643 SATA

$229.99



Dell R630 8SFF 2.1Ghz 16-Core 128GB H730 RAID 10GB RJ-45 NIC 2x750W PSU 8x Trays picture

Dell R630 8SFF 2.1Ghz 16-Core 128GB H730 RAID 10GB RJ-45 NIC 2x750W PSU 8x Trays

$425.04



Dell R740XD2 26LFF 3.6Ghz 8-C 384GB H730P MINI RAID 2x10G SFP+ NIC 2x1100W Rails picture

Dell R740XD2 26LFF 3.6Ghz 8-C 384GB H730P MINI RAID 2x10G SFP+ NIC 2x1100W Rails

$2998.08



Dell R740XD2 26LFF 2.5Ghz 20-C 64GB H730P MINI RAID 2x10G SFP+ NIC 2x1100W Rails picture

Dell R740XD2 26LFF 2.5Ghz 20-C 64GB H730P MINI RAID 2x10G SFP+ NIC 2x1100W Rails

$3500.08



Inspur LSI 9300-8i Raid Card 12Gbps HBA HDD Controller High Profile IT MODE picture

Inspur LSI 9300-8i Raid Card 12Gbps HBA HDD Controller High Profile IT MODE

$15.98



LSI MegaRAID 9361-8i 12Gb PCIe 8-Port SAS/SATA RAID 1Gb w/BBU/CacheVault/License picture

LSI MegaRAID 9361-8i 12Gb PCIe 8-Port SAS/SATA RAID 1Gb w/BBU/CacheVault/License

$39.95



LSI MegaRaid 9361-8i 12Gbps SAS / SATA Raid Controller PCIe x8 3.0 Tested picture

LSI MegaRaid 9361-8i 12Gbps SAS / SATA Raid Controller PCIe x8 3.0 Tested

$29.00



ORICO Multi Bay RAID Hard Drive Enclosure USB 3.0/ Type-C For 2.5/3.5'' HDD SSDs picture

ORICO Multi Bay RAID Hard Drive Enclosure USB 3.0/ Type-C For 2.5/3.5'' HDD SSDs

$87.99



4 Bay RAID External Hard Drive Enclosure for 2.5/3.5

4 Bay RAID External Hard Drive Enclosure for 2.5/3.5" SATA HDD/SSD

$79.99



Dell LSI 9440-8i MegaRAID 8 Port 12GB/s HBA High Profile RAID Controller YW3J6 picture

Dell LSI 9440-8i MegaRAID 8 Port 12GB/s HBA High Profile RAID Controller YW3J6

$84.99