[Cooker] [proposal] contrib process enhancement
OK, now that i have some free time to tr^H^Hdiscuss, here are some thought about how to make contributions easier and better. I propose to discuss here, and set up some page on the wiki when a consensus could be reached. 1) gpg signature of the contrib Currently, contrib package are not signed, or are individually signed by contribuer, making urmi yells every time you install them. I'd like to have them signed, and preferentially with a unique and official mdksoft key, instead of individual contributers keys. The reason if first they are official mdk packages first, it's easier to manage second (i don't want to add a new key to rpm database each time a new contributer arrive) 2) unique urpmi database on klama Currently, each time a package has some dependance to build, we have to first check if this dependance is either a contrib or a main package. This is both ugly and unpractical. I want to be able to install foo-devel with a single command, and let the system find where to fetch this package from 3) access to incoming Only Lenny is able to retrieve SRPM uploaded on ftp.mandrakesoft.com, so we generally ask people to also provide a link to retrieve it from elsewhere. All contributers should be able to access incoming, whatever location it is. Why not make people upload directly on klama ? 4) better mail adresses Each contributer should have some kind of official mandrake-linux mail adress, and be able to change real adress it correspond to easily. As a contributer is someone with a shell account on klama, why not use a SMTP there, delivering mail locally, making each one able to redirect mail to where he wants with a simple .forward ? 5) better mailing lists Do we really need two mailing-lists (maintainer and compil) ? Could thoses lists get subscribers only,so as to avoid spam ? Could subscription to those mailing list becomes automated (one shell account - one mail adress - one subscription) 6) anything else ? -- As soon as a still-to-be-finished computer task becomes a life-or-death situation, the power fails. -- Wood's Axiom
[Cooker]
[EMAIL PROTECTED] guillomovitch]$ rpmlint -i rpm/RPMS/i586/gkrellm-2.1.21-1mdk.i586.rpm [..] E: gkrellm only-non-binary-in-usr-lib There are only non binary files in /usr/lib so they should be in /usr/share. This rpmlint check should look for files exclusively, as the package only contains empty directories in /usr/lib: [EMAIL PROTECTED] guillomovitch]$ rpm -qlp rpm/RPMS/i586/gkrellm-2.1.21-1mdk.i586.rpm | grep /usr/lib /usr/lib/gkrellm2 /usr/lib/gkrellm2/plugins /usr/lib/menu/gkrellm BTW, why is /usr/lib/menu not /usr/share/menu ? -- Guillaume Rousse Inside every large program is a small program struggling to get out -- Thoreau's Theories of Adaption n°10
Re: [Cooker] perl-Magick relocation error
Ainsi parlait Gary Walsh : Thierry Vignaud wrote: Gary Walsh [EMAIL PROTECTED] writes: I am trying to run the BINS web photo album program and I get the following perl error: /usr/bin/perl: relocation error: /usr/lib/perl5/site_perl/5.8.1/i386-linux-thread-multi/auto/Image/Magick/ Magick.so: undefined symbol: InitializeMagick Are you sure you're using the mdk build of perl-Magick, and not the plf build ? I've seen similar problems already. What is a relocation error? the linker cannot relocate symbol = the package needs to be rebuild and relinked with current version of imagemagick libraries I have the latest rpm from cooker. There doesn't seem to be an SRPM on the cooker mirrors, so I can't rebuild it myself. It is part of the ImageMagick package. -- Any line, wire, or cloth cut to length will be too short -- Dimensions General Shefields Corollaries n°1
[Cooker] Re: kronolith or php-pear RPM in 9.2 Contribs missing php-pear Date module
This is better to discuss here... Ainsi parlait Colin Guthrie : Guillaume Rousse wrote: So it seems to be a packaging problem with the pear package, right ? I just had a look at the package, but i'm not php fluent enough to fix it, so i prefer to let its maintainer fix it. I guess it looks that way! There may have been a decision to stop including the Date code, perhaps PHP's own built in functions superceeded them at somepoint. I'll do a little digging and see if I can find out, otherwise, I'll talk to the maintainer and see what I can work out. Perhaps I'll redo the package myself and pass it on to him. However I was quite surprised to see this package doesn't carry its sources, but fetch them at buildtime. This break general packaging practice, and prevent offline building, so i don't see the point. Yeah, this does seem a little odd. I agree that it should not download them at build time (admitidly it's not technically building anything, seeing as this is PHP!!) for offline building. That said, with PHP, you're generally doing web stuff and are usually online, but this is still breaking the good practice rules! BTW, the url should also be changed to http://pear.php.net Cool. I'll have a fiddle myself and see what I can do. All the best. Col. -- ++ | Colin Guthrie | ++ | [EMAIL PROTECTED] | | http://colin.guthr.ie/ | ++ If at first you don't succeed, try management. -- All vacations and holidays create problems, except for one's own -- Murphy's Laws on Work n°34
Re: [Cooker] unable to use bewan drivers with 2.4.22-10mdk
Ainsi parlait Guillaume Rousse : Using latest bewan drivers release (7.4), the modules build correctly, but fails to install as depmod reports unresolved symbols. It works OK with 2.4.21-0.13mdk however. New problem today: the set of unresolved symbols changed. The only change was to remove the PCI card, as i need to keep it on the old gateway until i succeed building the drivers on the new one. Now it looks for unversioned symbols, whereas the default mdk kernel uses versioned ones... depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/atm/unicorn_pci_atm.o depmod: pci_write_config_byte depmod: send_sig depmod: flush_signals depmod: schedule_timeout depmod: __wake_up depmod: __kfree_skb depmod: alloc_skb depmod: __generic_copy_from_user depmod: vsprintf depmod: kmalloc depmod: pci_free_consistent depmod: pci_enable_device depmod: pci_disable_device depmod: boot_cpu_data depmod: cpu_raise_softirq depmod: pcibios_present depmod: pidhash depmod: free_irq depmod: __out_of_line_bug depmod: get_random_bytes depmod: iounmap depmod: pci_alloc_consistent depmod: interruptible_sleep_on_timeout depmod: __ioremap depmod: del_timer depmod: atm_charge depmod: mod_timer depmod: pci_release_regions depmod: kfree depmod: ___pskb_trim depmod: request_irq depmod: __verify_write depmod: exit_files depmod: skb_over_panic depmod: pci_set_master depmod: xtime depmod: pci_find_device depmod: sprintf depmod: daemonize depmod: jiffies depmod: softnet_data depmod: printk depmod: atm_dev_register depmod: add_timer depmod: complete_and_exit depmod: irq_stat depmod: shutdown_atm_dev depmod: kernel_thread depmod: __const_udelay depmod: __generic_copy_to_user depmod: pci_request_regions depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/atm/unicorn_usb_atm.o depmod: send_sig depmod: flush_signals depmod: schedule_timeout depmod: usb_ifnum_to_if depmod: __wake_up depmod: __kfree_skb depmod: alloc_skb depmod: __generic_copy_from_user depmod: vsprintf depmod: kmalloc depmod: usb_deregister depmod: boot_cpu_data depmod: cpu_raise_softirq depmod: pidhash depmod: usb_free_urb depmod: __out_of_line_bug depmod: get_random_bytes depmod: interruptible_sleep_on_timeout depmod: usb_alloc_urb depmod: usb_register depmod: del_timer depmod: atm_charge depmod: usb_set_configuration depmod: mod_timer depmod: kfree depmod: ___pskb_trim depmod: usb_string depmod: __verify_write depmod: exit_files depmod: skb_over_panic depmod: xtime depmod: usb_submit_urb depmod: usb_control_msg depmod: sprintf depmod: usb_get_current_frame_number depmod: daemonize depmod: jiffies depmod: timer_int_counter depmod: softnet_data depmod: printk depmod: atm_dev_register depmod: add_timer depmod: complete_and_exit depmod: irq_stat depmod: shutdown_atm_dev depmod: kernel_thread depmod: __const_udelay depmod: usb_set_interface depmod: __generic_copy_to_user depmod: usb_unlink_urb depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/net/unicorn_pci_eth.o depmod: pci_write_config_byte depmod: send_sig depmod: flush_signals depmod: eth_type_trans depmod: schedule_timeout depmod: __wake_up depmod: __kfree_skb depmod: alloc_skb depmod: __generic_copy_from_user depmod: pskb_expand_head depmod: vsprintf depmod: skb_under_panic depmod: kmalloc depmod: pci_free_consistent depmod: pci_enable_device depmod: create_proc_entry depmod: alloc_etherdev depmod: pci_disable_device depmod: cpu_raise_softirq depmod: pcibios_present depmod: pidhash depmod: free_irq depmod: unregister_netdev depmod: __out_of_line_bug depmod: get_random_bytes depmod: iounmap depmod: pci_alloc_consistent depmod: interruptible_sleep_on_timeout depmod: __ioremap depmod: proc_mkdir depmod: del_timer depmod: register_netdev depmod: mod_timer depmod: pci_release_regions depmod: kfree depmod: ___pskb_trim depmod: remove_proc_entry depmod: request_irq
Re: [Cooker] Re: kronolith or php-pear RPM in 9.2 Contribs missing php-pear Date module
Ainsi parlait Pascal Terjan : The point is to decide if this is usefull as pear already has commands to install/update packages. It is. We want a single packaging system, not one per language (perl, php, etc...). -- Guillaume Rousse You can go anywhere you want if you look serious and carry a clipboard -- Murphy's Laws on Work n°4
Re: [Cooker] Re: unable to use bewan drivers with 2.4.22-10mdk
Ainsi parlait Juan Quintela : It appears that your are compiling your drivers agaist a wrong source for some reason. Can you give one address for the drivers, please? You are right, i failed to notice it :-( However, i'd like to know why urpmi kernel-source on a fleshly installed 9.1 download a source different from the installed kernel... BTW, and just by curiosity, what could have caused the unresolved symbols to change ? -- Guillaume Rousse When dealing with the public try to look unimportant, They may go and find someone else to ask that stupid question to -- Murphy's Bush Fire Brigade Laws n°4
[Cooker] unable to use bewan drivers with 2.4.22-10mdk
Using latest bewan drivers release (7.4), the modules build correctly, but fails to install as depmod reports unresolved symbols. It works OK with 2.4.21-0.13mdk however. depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/atm/unicorn_pci_atm.o depmod: alloc_skb_R4de3cf22 depmod: shutdown_atm_dev_R38c2da4d depmod: skb_over_panic_Rc26dd3f9 depmod: irq_stat_R346aa8e2 depmod: exit_files_Red321d2d depmod: softnet_data_Ra505638d depmod: atm_dev_register_R29776e45 depmod: pidhash_R41eb4b03 depmod: flush_signals_Raa0f2644 depmod: __kfree_skb_R255859db depmod: ___pskb_trim_Rfc0bf7f6 depmod: send_sig_R14b22ade depmod: atm_charge_R7627be7b depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/atm/unicorn_usb_atm.o depmod: usb_ifnum_to_if_R6c4bbdf3 depmod: alloc_skb_R4de3cf22 depmod: usb_string_R5453acb3 depmod: shutdown_atm_dev_R38c2da4d depmod: skb_over_panic_Rc26dd3f9 depmod: usb_unlink_urb_R0f9211b4 depmod: irq_stat_R346aa8e2 depmod: exit_files_Red321d2d depmod: usb_free_urb_Rf2a6d6fb depmod: softnet_data_Ra505638d depmod: usb_alloc_urb_R0fbd4c7d depmod: atm_dev_register_R29776e45 depmod: usb_submit_urb_R4df3fd58 depmod: usb_register_Rde3a1aaa depmod: pidhash_R41eb4b03 depmod: usb_get_current_frame_number_R3222916c depmod: usb_deregister_R39c2e0c1 depmod: flush_signals_Raa0f2644 depmod: __kfree_skb_R255859db depmod: usb_set_configuration_R5fc86bda depmod: ___pskb_trim_Rfc0bf7f6 depmod: send_sig_R14b22ade depmod: usb_set_interface_R76c42b4e depmod: usb_control_msg_Ra8a99003 depmod: atm_charge_R7627be7b depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/net/unicorn_pci_eth.o depmod: alloc_skb_R4de3cf22 depmod: skb_over_panic_Rc26dd3f9 depmod: irq_stat_R346aa8e2 depmod: exit_files_Red321d2d depmod: softnet_data_Ra505638d depmod: remove_proc_entry_Rf05ffaa5 depmod: create_proc_entry_R74d25774 depmod: pidhash_R41eb4b03 depmod: eth_type_trans_Rebe24c2f depmod: netif_rx_R74e09bef depmod: proc_mkdir_Rd1a9779d depmod: flush_signals_Raa0f2644 depmod: pskb_expand_head_R9d6758a3 depmod: __kfree_skb_R255859db depmod: skb_under_panic_R9899c84a depmod: ___pskb_trim_Rfc0bf7f6 depmod: send_sig_R14b22ade depmod: *** Unresolved symbols in /lib/modules/2.4.22-10mdk/kernel/drivers/net/unicorn_usb_eth.o depmod: usb_ifnum_to_if_R6c4bbdf3 depmod: alloc_skb_R4de3cf22 depmod: usb_string_R5453acb3 depmod: skb_over_panic_Rc26dd3f9 depmod: usb_unlink_urb_R0f9211b4 depmod: irq_stat_R346aa8e2 depmod: exit_files_Red321d2d depmod: usb_free_urb_Rf2a6d6fb depmod: softnet_data_Ra505638d depmod: usb_alloc_urb_R0fbd4c7d depmod: usb_submit_urb_R4df3fd58 depmod: usb_register_Rde3a1aaa depmod: remove_proc_entry_Rf05ffaa5 depmod: create_proc_entry_R74d25774 depmod: pidhash_R41eb4b03 depmod: eth_type_trans_Rebe24c2f depmod: usb_get_current_frame_number_R3222916c depmod: netif_rx_R74e09bef depmod: proc_mkdir_Rd1a9779d depmod: usb_deregister_R39c2e0c1 depmod: flush_signals_Raa0f2644 depmod: pskb_expand_head_R9d6758a3 depmod: __kfree_skb_R255859db depmod: usb_set_configuration_R5fc86bda depmod: skb_under_panic_R9899c84a depmod: ___pskb_trim_Rfc0bf7f6 depmod: send_sig_R14b22ade depmod: usb_set_interface_R76c42b4e depmod: usb_control_msg_Ra8a99003 -- Guillaume Rousse If you can keep your head when all about you are losing theirs, maybe you just don't understand the situation -- Sad Truths of Life n°6
Re: [Cooker] Suggestion for Frozen Bubble
Ainsi parlait Guillaume Cottenceau : Yeah why not. Though if I really get time to spend on it I'll consider merging network stuff. [+] -- You know how most packages say Open here -- Why Why Why n°50
[Cooker] dia segfault when exporting with builtin eps converter
[EMAIL PROTECTED] presentation]$ dia --export-to-format=eps-builtin schema_general.dia - segfault [EMAIL PROTECTED] presentation]$ dia -e schema_general.eps schema_general.dia - works The first one worked yesterday, and i urgently need it :-) -- All warranty and guarantee clauses become void upon payment of the final invoice -- Financial General Shefields Corollaries n°2
Re: [Cooker] dia segfault when exporting with builtin eps converter
Ainsi parlait Götz Waschk : Am Freitag, 7. November 2003, 12:47:01 Uhr MET, schrieb Guillaume Rousse: [EMAIL PROTECTED] presentation]$ dia --export-to-format=eps-builtin schema_general.dia - segfault [EMAIL PROTECTED] presentation]$ dia -e schema_general.eps schema_general.dia - works The first one worked yesterday, and i urgently need it :-) There's no such format: -t, --export-to-format=FORMAT Export to file format and exit. Supported formats are: cgm, dia, dxf, eps, eps-pango, fig, mp, plt, hpgl, png, shape, svg, tex, wpg It also crashes if you specify something like eps-xxxbuildtin, it will always crash. BTW what's wrong with the output from the 2nd command line? See http://www.lysator.liu.se/~alla/dia/faq.html#ExportFormats Pango-randered eps is ugly when converted to pdf. -- The buddy system is essential to your survival; it gives the enemy somebody else to shoot at -- Murphy's Military Laws n°12
Re: [Cooker] reliable mirrors . . .
Ainsi parlait Stefan van der Eijk : True, cause you don't have access to what should be your reference point. I'd really like to have someting as RSS to publish mirrors list. Mandrake would publish it this way, third-party projects as PLF or JPackage would also, and tools as urpmi.setup could easily retrieve and use them... -- Guillaume Rousse There is no remedy for sex but more sex -- Murphy's Laws on Sex n°6
Re: [Cooker] Re: [CHRPM] xscreensaver-4.14-1mdk
Ainsi parlait Götz Waschk : Am Dienstag, 4. November 2003, 22:23:53 Uhr MET, schrieb Christian Bricart: -%define enable_xmatrix 0 -%define enable_extrusion 0 +%define enable_xmatrix 1 +%define enable_extrusion 1 Oops..? xmatrix is back in main? No license conflict anymore..? Or is it just in for the premiere of Revolutions tomorrow..? :) Oh fuck, I haven't changed my private package back for the Cooker release. I'll remove these two in the 2mdk release. Could you also add an additional plf switch, triggering all of these individual ones, and provide a plf build ? -- Substitutions never taste right -- Murphy's Food Laws n°6
Re: [Cooker] Re: [CHRPM] xscreensaver-4.14-1mdk
Ainsi parlait Götz Waschk : Am Mittwoch, 5. November 2003, 10:12:25 Uhr MET, schrieb Guillaume Rousse: Could you also add an additional plf switch, triggering all of these individual ones, and provide a plf build ? What would you prefer, an additional subpackage for extrusion and the 2 matrix savers or a plf package of xscreensaver with these included? Good question. So far we've used both solutions (avifile vs mplayer), mailny because there wasn't any other solution for mplayer. Maybe it's time to discuss this choice, and chosse an overall policy. -- Your side of the tent is the side that leaks -- Murphy's Laws of Camping n°14
Re: [Cooker] Re: [CHRPM] urpmi-4.4-42mdk
Ainsi parlait Thierry Vignaud : Guillaume Rousse [EMAIL PROTECTED] writes: Guillaume Rousse [EMAIL PROTECTED] 4.4-42mdk - added bash-completion - spec cleanup - bziped additional sources this package and especially its spec file is in cvs. i just committed your changes in cvs but do not forget this when altering packages whose upstream maintainer is mandrake and ask someone to commit them back into cvs the spec file is automatically commited to package CVS, why duplicate it in another CVS ? However, the shell source should be commited, the only problem was to know who to ask, as there is no maintainer anymore for urpmi :-( -- Trying to encapsulate your world of personal computing into an 8 1/2 x 11 notebook computer is like attending a special seminar at Murphy's Institute. -- Murphy's Law of Miniaturization
Re: [Cooker] Re: [CHRPM] urpmi-4.4-42mdk
Ainsi parlait Andrey Borzenkov : Guillaume Rousse [EMAIL PROTECTED] 4.4-42mdk - added bash-completion - spec cleanup - bziped additional sources hey, if we start to ship completion functions with packages what about Zsh? Well, i don't know if situation is the same. bash is maintained independently from bash-completion project, wich is a collection of shell function for completing common unix command. So bash-completion is package separatly from bash itself first, from commands it provide completion to also. It often turns into versioning nightmare: see how different bash version are handled into bash-completion, for instance. My own point is that commands that are mandrake specific (urpmi, rpmmon, net-profile, etc...) are not supposed to be installed differently that from mandrake packages, so it is better to ship them this way, so as to solve this versioning issue first, than contribute them to distribution-independant bash-completion project. On the other hands, completion for other tools (lilo, freeciv, dsniff, to cite a few from my own) are better handled at bash-completion project itself, so as to be usable for other distributions also. In ideal world, they should even get managed by upstream developpers. How does that translate to zsh ? BTW, feel free to use the cached package list (/var/lib/urpmi/names.*) fpons implemented at my demand to speed up zsh completion also fro urpmi. -- Murphy's Law will go into effect at the beginning of an evaluation. -- Murphy's Laws of Teaching n18
[Cooker] Re: [Plf-discuss] Fortune
Ainsi parlait Götz Waschk : Am Montag, 20. Oktober 2003, 16:13:40 Uhr MET, schrieb Olivier Thauvin: Le Lundi 20 Octobre 2003 16:07, Joao S. O. Bueno a écrit : There are a lot of Fortune extras on PLF. However, it seems like fortune itself is missing from MAIN in mdk9.2. Is it's place in contrib? Well, if it is, either way, there is no contrib for 9.2 I can find yet. It's not in contrib but in the main distribution. It's even on the 9.2 CDs (2nd CD in my PowerPack), but the package is named fortune-mod. I think this package would gain being split between engine (fortune binary) and several data files. Moreover, there should be done something about line-wrapping, such as defining a standard character/line limit (72) and enforcing it someway. I'm thinking of adding a perl script in fortune package, and using it at buildtime to correct fortune data files before building binary indexes. -- Guillaume Rousse Scout troops of the opposite sex are only encountered while dealing with the diarrhea mentioned above -- Murphy's Laws of Camping n°5
Re: [Cooker] The JPackage Project.
Ainsi parlait Ron Stodden : But nowhere from MandrakeSoft do I see any indication whether 9.2 as installed includes Java support and if not what are the recommended RPMs to add Java support. There is a wiki available for the communauty. Feel free to supply any necessary explanation there instead of whining after MandrakeSoft. What is the modus operandum necessary for us to make use of appropriate ones of these RPMs to achieve Java operability in Mandrake 9.2? Whats' it all about, please? Is this directory a new medium (NOT ?media?! ugh!) for urpmi? If so, what do we install? check the url I suspect the answer should have been ¨Yes¨, but none of the jpackage RPMs include any summary text in their RPM, so how are we to know what each is and whether we might need it? [EMAIL PROTECTED] RPMS.free]$ LANGUAGE=C rpm -qip ant-1.5.4-2jpp.noarch.rpm Name: ant Relocations: (not relocateable) Version : 1.5.4 Vendor: JPackage Project Release : 2jpp Build Date: mer 13 aoû 2003 12:23:42 CEST Install date: (not installed) Build Host: shitake.truemesh.com Group : Development/Build Tools Source RPM: ant-1.5.4-2jpp.src.rpm Size: 846189 License: Apache Software License Packager: Paul Nasrat [EMAIL PROTECTED] URL : http://ant.apache.org/ Summary : Ant build tool for java Description : Ant is a platform-independent build tool for java. It's used by apache jakarta and xml projects. BTW, if you don't know what is it, you don't need it... -- Guillaume Rousse Machines that have broken down will work perfectly when the repairman arrives -- Murphy's Laws on Work n°31
Re: [Cooker] The JPackage Project.
Ainsi parlait guran : söndagen den 19 oktober 2003 13.34 skrev Guillaume Rousse: Ainsi parlait Ron Stodden : But nowhere from MandrakeSoft do I see any indication whether 9.2 as installed includes Java support and if not what are the recommended RPMs to add Java support. There is a wiki available for the communauty. Feel free to supply any necessary explanation there instead of whining after MandrakeSoft. Will you please give the URL and try to explain what is needed to belong to the communauty. To post on cooker is usually enough to claim being part of mandrake communauty :-) For the wiki, it is available at http://qa.mandrakesoft.com/wiki, and you need to ask an account to warly if you want to contribute. -- Guillaume Rousse A man can be happy with any woman as long as he doesn't love her -- Murphy's Laws on Sex n°46
Re: [Cooker] The JPackage Project.
Ainsi parlait John Allen : Ron Stodden wrote: Guillaume Rousse wrote: [EMAIL PROTECTED] RPMS.free]$ LANGUAGE=C rpm -qip ant-1.5.4-2jpp.noarch.rpm Name: ant Relocations: (not relocateable) [snipped] Thanks! I will investigate ant some more.The 9.2 download edition installed jaffe for me ... You can get the Sun j2sdk/j2re from MandrakeClub Unfortunatly, it is not jpakage aware, meaning alternatives won't be there. I hope in the future we should be able to merge those packages, but currently, i'd rather advise to build jdk packages from jpp nosrc.rpm. -- Guillaume Rousse If it moves, salute it; if it doesn't move, pick it up; if you can't pick it up, paint it -- Murphy's Military Laws n°4
Re: [Cooker] 9.2 disasters list (continuing)
Ainsi parlait Ron Stodden : Buchan Milne wrote: So, only super experts are able to use the rescue option. Oh, I forgot, you missed that option. I doubt that rescue will run if / is full. rescue option from install cdrom. -- Guillaume Rousse If it moves, salute it; if it doesn't move, pick it up; if you can't pick it up, paint it -- Murphy's Military Laws n°4
Re: [Cooker] errata: kernel-source not on 9.2 CDs?
Ainsi parlait [EMAIL PROTECTED] : I cannot say i am very happy with this descission. Or was it a mistake? I surely hope so. Explaining people they have to install correct kernelsource to avoid problems when using/installing nvidia, win4lin, vmware and a dozen other things is difficult enough, not putting them on the download edition is not going to make it better. If it was a space consideration, why do we have 2 apache-source (what is it for?) and 3 (!) versions of the kernel docs there (and probably lots of other redundant stuff). Which remember me that if we were able to split headers from real sources, you won't need to install 30Mo package to build those additional kernel modules. Maybe there are technical arguments i didn't caught however, as it seems we already had this discussion. -- Guillaume Rousse The night before the English history midterm, your Biology instructor will assign two hundred pages on planaria. Corollary: Every instructor assumes that you have nothing else to do except study for that instructor's course. -- Laws of Applied Terror n°4
[Cooker] Re: [CHRPM] mplayer-0.91-8mdk
Ainsi parlait Gwenole Beauchesne : Note: If you want to play Real content, you need to have the content of RealPlayer's Codecs directory in /usr/lib64/RealPlayer8/Codecs Hmmm, there is a PLF real codecs package now, it would make more sense to add the amd64 codecs there and drop this additional paragraph. This PLF build has additional support for AAC decoding with libfaad, which is covered by software patents. It also includes support for reading DVDs encrypted with CSS which might be illegal in some countries. Is this really wise to have this paragraph available even for standard build? Sure it's more useful, but it's a bit too much sensible. It would be safer to have it only in PLF build IMHO. -- There is nothing more satisfying that having someone take a shot at you, and miss -- Murphy's Military Laws n°97
Re: [Cooker] Re: [Contrib-Rpm] anthill-0.2.4-2mdk
Ainsi parlait Vincent Danen :
On Mon Oct 06, 2003 at 07:48:29PM +0200, Guillaume Rousse wrote:
Oden.. nice to see, but you didn't install it in a good way.
You have include/ exposed, which would have been fine for 0.2.3 or
earlier, but the layout should really be something like:
/var/www/anthill
rather than /var/www/html/anthill. Then you just expose
/var/www/anthill/html (ie. via an Alias or a symlink), but you keep
include/, etc/, etc. unexposed and entirely unreachable for maximum
security.
There have been some discussion previously about this, see
http://qa.mandrakesoft.com/twiki/bin/view/Main/PackagingTask#Web_applicat
ions for kind of synthesis.
The point was to use /var/www/html/%{name} for every application, and to
use FHS compliant location for non-web files. If you have a configuration
directory for anthill, it seems for me more logical to use /etc/anthill
for it than /var/www/anthill/etc, for instance. The same could be said
for include, that should rather go into /usr/share/anthill.
I can fix this a little later on if you like (or you can). I'm not on
the cooker list anymore so you'll have to cc me.
I'd prefer to restart this discussion on web applications policy first...
Ack... this is why I don't think web apps should be rpm packaged.
And this is the reason why i think web apps installation should be discussed
first before being packaged :-)
Can you make it even more convoluted and hard for people to use?
That's the same answer i had from java developpers when i tried to explain
them putting all depended libs and build binaries in CVS was wrong.
Wouldn't it make more sense to have something like:
/var/www/packages/%{name}
that you install everything into and then add a Directory/Directory
clause to httpd.conf or some other include file so we can alias, for
example
/geeklog/ /var/www/html/packages/geeklog/public_html/
or
/anthill/ /var/www/html/packages/anthill/html/
so we don't screw around with the normal way of doing things?
If the normal way is plainly wrong because developpers have no clues about
system administration, for whatever reason, should we just keep it this way ?
Web apps are especially concerned because they are mostly using Perl, PHP, and
other multiplatform languages, meaning:
- some developpers come from windows world
- developpers have to take care about windows world
As for Java software, the result is to bring all the default from broken
platforms (meaning win32) to Unix world.
Think of a native, purely unix application, where the author made a wrong
installation decision. As a packager, it is yours responsability to make it
better. This can be achieved in multiple way, such as patching, using
symlink, etc... See how apache, for instance, is itself packaged whereas
standard install drop everything under a single TLD.
You start throwing anthill/include into /usr/share/anthill and anthill/etc
into /etc/anthill and you're going to be messing up all kinds of people
unless you plan on rewriting the docs. This goes for every web app, not
just anthill.
And all other applications too, where packager work is usually
underdocumented.
Anways anthill/etc is not a configuration directory.. it has upgrade files,
etc. The configuration is in anthill/include. So you propose to put
anthill/include into /etc/anthill and anthill/etc into /usr/share/anthill?
I propose to avoid the windows-like habit of throwing everything under a
single directory, without ever considering if it makes sense to put README,
LICENSE and .po files there...
I don't however have definitive answers about where to put everything else,
explaining why i'm proposing to discuss it.
For configuration files, i find it more logical as a sysadmin point of view
(we are discussing client-server applications, right) to find configuration
files under /etc, rather to have to run rpm -q --docfiles to see where they
are installed. And i find more secure to install something out of web root,
instead of under web root with access denied.
If you do stuff like that, authors and people on mailing lists will start
people to avoid Mandrake packaged web apps like the plague... you've just
made support 3x as difficult because the authors aren't going to know where
stuff goes.
Until someone realise than using just urpmi imp is enough to install
everything needed in a secure manner, without useless aditional craps, etc.
Or that urpmi bugzilla install database, cron task, etc... automagically.
This is a general packaging problem: either you favor respect for original
software, and let users do everything at hand (slackware style), or you favor
integration and distribution coherency, and provide user a framework for
repeated tasks (Debian style). I am personaly much in favour of 2nd one,
provided we could _document_ the additonal added value, which is clearly
lacking currently :-(
You'll also likely screw up
Re: [Cooker] Re: [Contrib-Rpm] anthill-0.2.4-2mdk
Ainsi parlait Vincent Danen :
On Fri Oct 03, 2003 at 01:01:07AM +0200, Oden Eriksson wrote:
Oden.. nice to see, but you didn't install it in a good way.
You have include/ exposed, which would have been fine for 0.2.3 or earlier,
but the layout should really be something like:
/var/www/anthill
rather than /var/www/html/anthill. Then you just expose
/var/www/anthill/html (ie. via an Alias or a symlink), but you keep
include/, etc/, etc. unexposed and entirely unreachable for maximum
security.
There have been some discussion previously about this, see
http://qa.mandrakesoft.com/twiki/bin/view/Main/PackagingTask#Web_applications
for kind of synthesis.
The point was to use /var/www/html/%{name} for every application, and to use
FHS compliant location for non-web files. If you have a configuration
directory for anthill, it seems for me more logical to use /etc/anthill for
it than /var/www/anthill/etc, for instance. The same could be said for
include, that should rather go into /usr/share/anthill.
I can fix this a little later on if you like (or you can). I'm not on the
cooker list anymore so you'll have to cc me.
I'd prefer to restart this discussion on web applications policy first...
--
Guillaume Rousse
If they're just accessories, how come we want them so badly?
-- Murphy's Law of Accessories
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : On Thu Oct 02, 2003 at 12:47:50AM +0200, Buchan Milne wrote: I don't believe there is anyone enjoying having to mantain 4 different packages of the same software when one would suffice. Probably not. But if you, as a contributor, have a cooker machine and compile on cooker, how can you possibly know if your package, despite having conditional build macros, will work with an older distrib if you don't take the time to build and test on that old platform? So you *do* need to maintain it in such a manner otherwise you're just pumping out stuff that pretends to work on old distribs and you really don't have a clue if it does or not. But, the point is that if people are interested in maintaining their packages for rebuild on older releases, then it may be possible to make this easier by having automated rebuilds on stable releases. Yes, but don't you understand that automatic rebuilds is not enough? It needs to be *tested* first. I didn't pretend it was enough, just that getting stable versions for free, without any additional human resources need, would be a gain over current situation. It would not bring an updated version of stable distribution for sure, but at least provide packages with far better chance of working on stable version that cooker packages. And nothing prevent packagers from testing on stable, as well as they are testing on cooker. -- The Cavalry doesn't always come to the rescue -- Murphy's Military Laws n°76
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Luca Berra : Would other contributors be adverse on taking the commitment of maintaining 'stable' packages for 'supported' version of mandrake (that should be 18 months or 3 releases IIRC)? Yes, unless someone give me the means to test those packages. I only run cooker for my desktops, and stable for my servers. I don't want to maitain three other versions just to test packages. Morevoer, the more version you want to support, the more difficult it is. Given the current mdk policy of releasing often, and the current situation of no support at all (security excepted), supporting lastest stable version only seems for me an achievable goal, not three. -- Any cool program always requires more memory than you have. -- Murphy's Computer Laws n°2
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : Actually, there is no point just explaining servers in contrib are not updated. A real explanation of mandrake policy would be far better: - what is main, what is contrib, and what is update ? - what does get updated ? etc... Good grief.. everyone wants policy policy policy... =) Main is what is in the main tree. You know, what comes in the download edition. The stuff that is labelled not contribs. The meat of the distro. The equivilant of Mandrake-devel/cooker/i586/Mandrake/RPMS Contribs is contribs. It's labelled as such. updates? Should be pretty obvious. Updates are stuff in main that need fixing for whatever reason. Main gets updated. contribs doesn't. I don't see what is so difficult about this. This is how it has *always* been. It hasn't changed. No need for a policy regarding it. This is just how it's done. I know all of this, just because i've been there for 4 years now. My point is that i'm fed up with oral tradition. No one knows about contribs outside cooker communauty. Among the few newbies that know about urpmi, they always use urpmi.update whereas none of their sources ever change. Etc... What we lack here is a clear explanation of mdk functionning targeted at outside world. -- If your attack is going really well, it's an ambush -- Murphy's Military Laws n°47
Re: [Cooker] And next ?
Ainsi parlait Vincent Danen : really nasty bugs for fixing... not little things like fonts in gkrellm). Vincent, yet another provocation, and i'll give you gkrellm package back :-) -- The legibility of the signature is inversely proportional to the importance of the artist. -- Caba's Axiom Concerning Signatures and Paintings
Re: [Cooker] And next ?
Ainsi parlait Buchan Milne : Vincent Danen wrote: On Mon Sep 29, 2003 at 11:37:22PM +0200, Buchan Milne wrote: This is what @linux-mandrake.com addresses are for. But again, there is no documented policy (and maybe even the resitrictions on what you should use it for) anywhere, so people don't know how to get one. Hasn't it always been to email Lenny and ask? And how reliable is this? What feedback do people get? What do you do if there is no reply? This is not a scalable solution ... And contributers not being subscribed to maintainers mailing-list, whereas half of traffic on this list is pure spam... This is clearly a point needing both a better explanation AND a technical solution. -- If you have your finger touching the rearview mirror that says objects in mirror are closer than they appear, how can that be possible? -- Why Why Why n°13
Re: [Cooker] And next ?
Ainsi parlait Buchan Milne : Brad Felmey wrote: On Tue, 2003-09-30 at 10:46, Michael Scherer wrote: On Tuesday 30 September 2003 05:38, Brad Felmey wrote: Also, a place where folks can go get urpmi lines. Not just for stuff like MdkClub and mirrors, but kind of like Debian has a list of misc. repositories available that make it easy to locate Mdk-specific packages for whatever the user is looking for. Almost all unsupported and unaffiliated, but at least a place where a user can go look for homegrown packages of stuff (a la Texstar, etc.). Do you mean something like http://plf.zarb.org/~nanardon/ No, that's only main, contrib, and plf. There are bunches more stuff out there. http://plf.zarb.org/~nanardon/?minor=1 (the equivalent of checking some box on the page) This should really be accessible from urpmi.org instead, and urpmi.org itself should get better organisezed, with more documentation. -- A filter set contains all imaginable wavelengths and widths- except the one combination you need -- Ralf's Laws of Observational Astronomy n°6
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : In my mind, the best solution is to put updates in Club. That way they make it to the Club mirrors and everyone can take advantage of them since non-Club members can also access them. That would be my best solution and my preference. Then I don't have to get involved at all. Unless that contributers and volonteers are two different set of people, with different practices, and sofar everything has been done to keep them separated. Even if i pesonally have an honorific club membership, i'd dislike having to ask yet another account just because i'd like to maintain my packages. -- Enough dirt will get tracked into the tent on the first day out, that you can grow the food you need for the rest of the trip in rows between sleeping bags -- Murphy's Laws of Camping n°22
Re: [Cooker] [Mandrake 10] An Idea for next MDK 10...
Ainsi parlait Vincent Danen :
BTW, I actually need a solution for this too. At present, all samba
builds from the source release get identical %{version}-%{release} which
is probably a bad thing ...
%define version 3.0
%define release 1
%define distro 92
Version: %version
Release: %release.%distromdk
so you end up with 3.0-1.92mdk
umm.. untested. =) But you get the idea.
Hardcoding the distribution target in your spec files prevent you from
rebuilding it for different distributions. I prefer Bucan's idea in this
regard.
Anyway, i don't think messing with tags having already a well-defined
semantics (such as release number or epoch) to workaround something not
supported by rpm itself is really a good idea. There is far too much risk to
break something for few added value IMHO.
--
There is nothing more satisfying that having someone take a shot at you, and
miss
-- Murphy's Military Laws n°97
Re: [Cooker] [Mandrake 10] An Idea for next MDK 10...
Ainsi parlait Austin : On 09/26/2003 09:38:37 AM, Götz Waschk wrote: It would be nice if packages would include an implicit epoch tag, so a package build on 9.2 will always be newer than one for 9.1 with the same version and release tags. Hmmm, this is a neat idea. At present, any mdk rpm you grab off the net, you have no idea what mandrake release it is built for. That's exactly why you shouldn't use individual packages, but only trust repositories. And this is a repository job to sort package by distribution target. -- Any manufacturer making his warranties dependent upon the device being earthed will only supply power cabling with two wires. -- Murphy's Hardware Laws n°4
[Cooker] Sorting improvement suggestions by categories
Every new release, there is the same question: what for the next version ? And everytime, this is the same never-ending thread mixing trivial suggestion to upgrade foo-0.1.1.1.1 to foo-0.1.1.1.2beta and utopic request for being taken account by management. So i'd like to propose a small, limited, simple policy about HOW to discuss distribution evolutions. First, simple package evolution. No need to ask, they have always been automatically upgraded. Don't worry, there will be foo-0.1.1.1.2beta as soon as it will be available. Second, simple feature addition. A simple mail should be enough. For instance, my own list for 9.2 was http://marc.theaimsgroup.com/?l=mandrake-cookerm=104889039014087w=2 Third, policy or functionning change. This is way more difficult to discuss, especially when appearing in a middle of a long, long thread. So i think those suggestion sould deserve their own thread, and use wiki as a way to offer a resume of the discussion once achieved, not for leading the discussion itself. The discussion about the mirror structures we already had falls in this category: http://marc.theaimsgroup.com/?l=mandrake-cookerm=106077548518257w=2 WDYT ? -- When you are in a hurry, the needle eye is always too small -- Murphy's Laws of Sewing n°5
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : On Wed Oct 01, 2003 at 01:50:24PM +0200, Guillaume Rousse wrote: In my mind, the best solution is to put updates in Club. That way they make it to the Club mirrors and everyone can take advantage of them since non-Club members can also access them. That would be my best solution and my preference. Then I don't have to get involved at all. Unless that contributers and volonteers are two different set of people, with different practices, and sofar everything has been done to keep them separated. Even if i pesonally have an honorific club membership, i'd dislike having to ask yet another account just because i'd like to maintain my packages. Fine. Instead of shooting down my suggestion, why not try to suggest one yourself? I don't have any, unfortunatly. However, i think we should make this distinction between contrib and club disapear. -- If you DO follow the advice of more experienced modemers and use a totally different password on every BBS you call, you will forget the password of the board where your date has left a message telling you where to meet them tonight -- Murphy's Laws of BBS'ing n°7
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Buchan Milne : Eric Fernandez wrote: Buchan Milne wrote: Which means that this information needs to be more readily accessible ... directly. Think about users who hardly have internet access ... I really think the documentation aspect (making it easy for newbies to find the documentation they need) is the problem that needs to be solved next. It is exactly my point, and pages like the one you did for your university mirrors are a good way. I shall try to do something graphical, then post it in my wiki (when Warly finally activates my access... Warly ? ;) ) If you mean that it's final place should be on a website/wiki, then IMHO http://mandrake.vmlinuz.ca is currently a better place (and you will get a password immediately). Even if no wiki is currently available, i really think we should better use domain name urpmi.org, and make it the mdk equivalent of apt-get.org. This domain name has been specificaly bought by Cedric for such purposes. So far it is just a bunch of already existing documentation, but it could largely get enhanced. Olivier's script currently known as a plf subproject http://plf.zarb.org/~nanardon should also be made available from there. Being independant from mdksoft would also save us from the legal concerns about direct links toward PLF co. -- Guillaume Rousse If you can't change the rules, then ignore them -- Anti- Murphy's Laws n°10
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : Suddenly I'm getting very tired of all of this again. Seems no matter how hard you try, someone has something negative to say without contributing something useful. Vincent, i didn't intend to be rude, just to say that we first to have to make the club work in a scalable and efficient manner first, and integrate it into standard development process whereas it is currently completly run in a parallel and amateurish way. Automatic rebuilding of all cooker packages for current stable version should be enough to solve contrib update problems IMHO. -- Guillaume Rousse Puritanism - The haunting fear that someone, somewhere, may be happy -- H.L. Mencken on Murphy n°7
Re: [Cooker] And next ?
Ainsi parlait Buchan Milne : On Wed, 1 Oct 2003, Levi Ramsey wrote: On Wed Oct 01 23:17 +0200, Luca Berra wrote: On Wed, Oct 01, 2003 at 01:42:40PM +0200, Guillaume Rousse wrote: And contributers not being subscribed to maintainers mailing-list, whereas half of traffic on this list is pure spam... and ppl actually discovering there is a maintainers mailing-list because they stumble upon a msg like this in cooker? Send an email with subscribe maintainers to [EMAIL PROTECTED], IIRC. (does this work? If so, should it be added to the wiki, or will this just worsen the problem below?) No, it should be done automatically by people in power to give someone contributer status, as part of welcome package: - klama account - bugzilla account - maintainer and compil subscription - @something.mandrake-linux.com adress We shouldn't have to require something. and BTW, maintainer should be subscribers-only. -- Guillaume Rousse The one item you need is always in short supply -- Murphy's Military Laws n°87
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Vincent Danen : It has been said clearly... many many many times. contribs is *entirely* unsupported. I don't know how many different ways I need to say the same thing. =) Just create one page on the wiki explaining it ? This way it could become the what are x and y sections page i'm asking for without finding the time to do it myself. -- Guillaume Rousse The world does not revolve on an axis -- Murphy's Laws on Sex n°52
Re: [Cooker] [Mandrake 10] Replacing proftpd by pureftpd ?
Ainsi parlait Han Boetes : Vincent Danen [EMAIL PROTECTED] wrote: On Thu Sep 25, 2003 at 07:59:30PM -0400, Levi Ramsey wrote: I think the policy should be that anything which listens on a port should not, under any circumstances, be in contribs, as contribs are not generally updated; I'm sure that someone will come along with a sendmail repository and do the updates themselves. I disagree. A lot of nice network-type software would be missing and I sure as heck don't want them all in main. What's wrong with having them in contribs? They aren't officially maintained... so what? Having them in contribs, joe sysadmin can grab the src.rpm for what he has installed, grab the new version or patch, and roll his own. It's still convenient for him to have it in contribs even if he doesn't get it via MandrakeUpdate. I does make sense if we have a note attached on them they can pose a security risk and if people use them they should take care they are up to date. Some sort of mechanism that keeps people informed about updates. For example xmule which was found to be exploitable recently. I expect more troubles from that program. A simple warning after installing wont even do... Can somebody come up with a decent solution to this problem? Actually, there is no point just explaining servers in contrib are not updated. A real explanation of mandrake policy would be far better: - what is main, what is contrib, and what is update ? - what does get updated ? etc... -- Never tell the Platoon Sergeant you have nothing to do -- Murphy's Military Laws n°63
Re: [Cooker] Xmule
Ainsi parlait Götz Waschk : Am Freitag, 26. September 2003, 13:02:25 Uhr MET, schrieb Michael Lothian: I know Xmule isn't offically part of Mandrake and only available from plf but... PLF has it's own list: [EMAIL PROTECTED] And has _nothing_ to do with mdk, as everyone knows :-P -- The probability of a diagram or a specification being omitted from a shop manual is directly proportional to its importance -- Mechanical General Shefields Corollaries n°2
Re: [Cooker] How to upgrade from 9.1 to cooker?
Ainsi parlait Nicholas Brown : Is this the correct way to upgrade from using 9.1 to using cooker? (are other steps needed etc?) (existing media are; main,contrib,plf,texstar) urpmi.removemedia -a urpmi.addmedia cooker-main ftp://ftp.sunet.se/pub/Linux/distributions/mandrake-devel/cooker/i586/Mandr ake/RPMS with ../base/hdlist.cz urpmi.addmedia cooker-contrib ftp://ftp.sunet.se/pub/Linux/distributions/mandrake-devel/contrib/i586 with ../../cooker/i586/Mandrake/base/hdlist2.cz urpmi.addmedia cooker-plf ftp://ftp.club-internet.fr/pub/linux/plf/cooker with hdlist.cz urpmi urpmi urpmi --auto-select urpmi kernel Safer bet: - urpmi urpmi before removing your medias, to get sure you have latest available version for your distro - urpmi urpmi before urpmi --auto-select, to upgrade it first -- The only thing more accurate than incoming enemy fire is incoming friendly fire -- Murphy's Military Laws n°88
Re: [Cooker] 9.2 ISOs has been sent
Ainsi parlait Teletchéa Stéphane : 2 - second i really dislike being held by the company AFTER the job is done. I hope Mandrake will tell for next release what then plan to do BEFORE we go on beta and RC. It is not really about what they are doing now, i perfectly understand the point, but i would like to know it before, like this i don't feel caught by then : when you expect a product by the end of september and it arrives by the end of october it is not the same thing ! If you've been following cooker from a long time, you should be able to see mandrakesoft has a very long established tradition of taking decision first, then eventually explain them thereafter, including to their own emplyees. Just because most developpers usually request cooker advices on technical decision doesn't means management feels compelled to do the same about corporate decisions... -- There is nothing more satisfying that having someone take a shot at you, and miss -- Murphy's Military Laws n°121
Re: [Cooker] 9.2 ISOs has been sent
Ainsi parlait Guillaume Cottenceau : Guillaume Rousse [EMAIL PROTECTED] writes: If you've been following cooker from a long time, you should be able to see mandrakesoft has a very long established tradition of taking decision first, then eventually explain them thereafter, including to their own emplyees. Btw, that's how most companies work, I think Sure. People tend to forget it, and expect mandrakesoft to behave as a nonprofit organisation. (even if I have no experience in other companies yet ;p). Soon, my dear, soon -- Program complexity grows until it exceeds the capability of the programmer who must maintain it -- Thoreau's Theories of Adaption n°8
Re: [Cooker] 9.2 ISOs has been sent
Ainsi parlait Michael Lothian : Me used to be a stock controller before I became a checkout bunny ;-) As the playboy bunnies ? You're walking half-naked in the store to promote computers ? -- Never stand when you can sit, never sit when you can lie down, never stay awake when you can sleep -- Murphy's Bush Fire Brigade Laws n°12
Re: [Cooker] ALSA fix
Ainsi parlait Adam Williamson : On Tue, 2003-09-23 at 09:01, Eric Fernandez wrote: Adam Williamson wrote: I think it's pretty hard to pick which should be the default. Even if it ought to be ALSA, it's by no means as clear-cut a decision as this particular user's experience indicates. Except if the installer offered the possibility to test audio at install : first with OSS driver, then with Alsa, asking the user to say if he hear a sound. WDYT ? It could be a nice feature for the 10. I don't like it. In most cases, it would be a needless extra step, and could well cause confusion. I think the current ploy of just trying to get the right default for each piece of hardware is correct. Add OSS is likely to die soon, so why waste effort to support it ? -- You can lead a man to slaughter, but you can't make him think -- Sad Truths of Life n°13
Re: [Cooker] Cooperation with RedHat Linux project?
Ainsi parlait Adam Williamson : On Mon, 2003-09-22 at 01:06, Pierre Jarillon wrote: Le Dimanche 21 Septembre 2003 16:41, David Walser a écrit : http://rhl.redhat.com/ From reading this, it sounds like the possibility of some cooperation between Cooker and the new RHL project is even more possible. RedHat said: We are excited to announce that we are working on an alliance with another well-known provider of Red-Hat compatible packages. I try to guess who is this well-known provider of Red-Hat compatible packages. IMHO, not so many ! Sounds like Freshrpms to me. I'd say fedora rather. -- In any given price estimate, final cost will exceed estimates by a factor of 2 -- Financial General Shefields Corollaries n°1
Re: [Cooker] Cooperation with RedHat Linux project?
Ainsi parlait Michael Scherer : RedHat said: We are excited to announce that we are working on an alliance with another well-known provider of Red-Hat compatible packages. I try to guess who is this well-known provider of Red-Hat compatible packages. IMHO, not so many ! Sounds like Freshrpms to me. what about fedora ? ( http://fedora.us ) Won: http://fedora.redhat.com/ -- If a program is useful, it will have to be changed -- Thoreau's Theories of Adaption n°4
Re: [Cooker] rxvt menu entry?
Ainsi parlait andre : Since when can mcc be run by non-root? Since there is a wrapper. -- Mother said there would be days like this, but she never said there would be so many -- Murphy's Laws on Work n°10
Re: [Cooker] rxvt menu entry?
Ainsi parlait [EMAIL PROTECTED] : I plead for some sanity, please put the menu entry back in. Let the user that doesn't want a terminal remove the entry (it would give them an excuse to learn menudrake), and don't make the 99.99% of us who do want it have to load up menudrake just so we can gain access to a terminal. How about a compromise? Put it in the Configuration menu (where it belongs). User terminals (gnome-terminal, konsole etc etc, can go in Terminals, but rxvt should only be needed for configuration work so that's where it belongs). And what about clearly labeling packages for which it is useless to ask anything to maintainers, as they know better than anyone else what is good for users, and they don't give a fuck about contributers ? A new private section, similar to main and contrib, could host all those packages, safe from hostile remarks and angry comments. -- Guillaume Rousse Inside every complex and unworkable program is a useful routine stuggling to be free -- DSP Paradox (Nolans Placebo) n°7
Re: [Cooker] make xconfig fails!
Ainsi parlait Robert Fox : After a fresh install and fresh kernel-source package - i tried to run a make xconfig and got this: -- [EMAIL PROTECTED] linux]# make xconfig rm -f include/asm ( cd include ; ln -sf asm-i386 asm) if [ -f .need_mrproper ]; then \ rm .need_mrproper; \ make mrproper; \ make preconfig; \ fi make -C scripts kconfig.tk make[1]: Entering directory `/usr/src/linux-2.4.22-6mdk/scripts' gcc -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -c -o tkparse.o tkparse.c gcc -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -c -o tkcond.o tkcond.c gcc -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer -c -o tkgen.o tkgen.c gcc -o tkparse tkparse.o tkcond.o tkgen.o cat header.tk ./kconfig.tk ./tkparse ../arch/i386/config.in kconfig.tk 3rdparty/lufs/Config.in: 2: unknown command make[1]: *** [kconfig.tk] Error 1 make[1]: Leaving directory `/usr/src/linux-2.4.22-6mdk/scripts' make: *** [xconfig] Error 2 Quick fix: edit 3rdparty/Config.in, and remove lufs reference. BTW, this kind of error appears very often. I really think we should find some generic way of testing taht xconfig works that could be included in the kernel spec. -- Never tell the Captain or Deputy Captain you have nothing to do -- Murphy's Military Laws n°61
Re: [Cooker] Upgrade from 9.1
Ainsi parlait Robin Rosenberg : söndagen den 14 september 2003 08.07 skrev Mike Eheler: Has anyone tried upgrading some installations that maybe have been worked in a bit? Some self-compiled programs, some mixed rpms, some external rpms (i.e. freshrpms, ximian, etc). freshrpms ? Mathhias only provides redhat rpms... -- Guillaume Rousse When the pin is pulled, Mr Grenade is not your friend. -- Murphy's New Military Laws n°1
Re: [Cooker] um.
Ainsi parlait Jan Ciger : Guillaume Cottenceau wrote: | I hate to say that, but I agree. But I let you also know that we | also have internal communications problems, and I also learnt | from linuxfr.org the presence of the mandrake page with the price | of ads :/. Wow, if even Mandrakesoft's employees didn't know, I guess the problems are much worse than just one botched press release. I hope you manage to fix whatever is broken, because this fracas didn't contribute to good image of Mandrake Linux and Mandrakesoft. Maybe the image shouldn't mask the reality too much. Mandrakesoft is a standard corporation, with bosses and employees. The communauty involvement has been made possible by some of these employees, and we all know that they are other developpers that don't give a fuck about us. And the real decision process stays in management hand anyway. I can't say I enjoy the situation, but I find whining every time this is clearly demonstrated a bit... naïve ? -- Guillaume Rousse Any circuit design must contain at least one part which is obsolete, two parts which are unobtainable and three parts which are still under development -- Murphy's Laws on Technology n°23
Re: [Cooker] Re: [Contrib-Rpm] sgrotum-1.2.6-1mdk
Ainsi parlait Pascal Terjan : Thierry Vignaud wrote: Han Boetes [EMAIL PROTECTED] writes: -- __ Men often believe -- or pretend -- that the Law .,-;-;-,. /'_\ is something sacred, or at least a science -- an _/_/_/_|_\_\) / unfounded assumption very convenient to '-=/\ governments. jgs `/_//_/-'\_\ is this really needed in package description ? I agree, putting a sample is nice, but a neutral one for most people would be better. You mean changing the turtle would be better, right :-) ? -- In a mummy bag the urgency of ones need to urinate is inversely proportional to the amount of clothing worn -- Murphy's Laws of Camping n°7
Re: [Cooker] Re: [Contrib-Rpm] sgrotum-1.2.6-1mdk
Ainsi parlait Thierry Vignaud : packages description is not places for sigs. Sure, changelog is way better for this, since you can put several sigs there :-) -- Tell a man you saw a flying saucer last night, and he'll believe you. Tell him your mechanical bank is in working order, and he'll put a coin in to see for himself. -- McCarthy's Axiom
Re: [Cooker] Re: [Contrib-Rpm] sgrotum-1.2.6-1mdk
Ainsi parlait [EMAIL PROTECTED] : Guillaume Rousse [EMAIL PROTECTED] said: Ainsi parlait Pascal Terjan : Thierry Vignaud wrote: Han Boetes [EMAIL PROTECTED] writes: -- __ Men often believe -- or pretend -- that the Law .,-;-;-,. /'_\ is something sacred, or at least a science -- an _/_/_/_|_\_\) / unfounded assumption very convenient to '-=/\ governments. jgs `/_//_/-'\_\ is this really needed in package description ? I agree, putting a sample is nice, but a neutral one for most people would be better. You mean changing the turtle would be better, right :-) ? Put a dog, or a bear. Or childrens (childreeens!) And some disclaimer: no animal has been harmed during the packaging of this software. -- The only time suppressive fire works is when it is used on abandoned positions -- Murphy's Military Laws n°16
Re: [Cooker] urpmi / apt4rpm - not for 9.2
Ainsi parlait Pierre Jarillon : Le Vendredi 12 Septembre 2003 17:07, Michael Scherer a écrit : It would be nice to make a package for Mandrake and Debian with the same tool. what would be the benefit ? To reduce the amount of time needed and spent to make packages. You're kidding ? Have you ever packaged anything ? Supporting multiple versions is already a nightmare on one single distribution. Supporting several distribution is even worst. And i my experience was about noarch package for mdk/redhat only. I can't imagine what i could be for native packages on distributions as different as Debian and Mandrake. being able to install the same package on the 2 distros ? Why not ? This is a good aim. Package are not tarball. The more general you are, the less added value you provide. I don't know the debian policy, but i know that our binary are not fully working on debian, and that mandrake naming scheme is different from the debian ones, so, dependancy will be unsatisfied. This is for LSB... And a long job... LSB is not supposed to make sure the C++ compiler is the same on both distro. You can't share C++ software if they're not build with the same compiler. LSB isn't supposed either to enforce the rpm version used. You can't share package if they use too different rpm format. LSB won't enforce neither compiler flags, optimisations policies, man page compression. Or if they do, they'll use the less constraining ones. LSB is mostly pushed by commercial companies that want to produce their package themselves, so as to be able to reach the maximum audience, without distributing their source to third-party packagers. You can't wait for miracles. So, maybe you think of something which can produce the files neeeded to produce a package for each distro, which would be quite interesting, but, not so useful, IMHO. This was my idea. Your opinion is interesting. If you're interested in xml, see what was done in JPackage project to produce redhat and mandrake spec from the same base XML file. http://www.zarb.org/horde/chora/cvs.php/xml-spec?rt=jpackageHorde=37124a6d30cd6b3e1288db79a6aefda6 -- Guillaume Rousse The probability of a hard-disk crash increases with the number of days since the drive was last backed up. -- Murphy's Law of Data Loss
Re: [Cooker] Re: [Contrib-Rpm] tvtime-0.9.10-1mdk
Ainsi parlait Austin : On 09/12/2003 06:30:49 PM, Guillaume Rousse wrote: [Contrib-RPM] -=-=-=- Name: tvtime Relocations: (not relocateable) Version : 0.9.10Vendor: MandrakeSoft Release : 1mdk Build Date: Sat Sep 13 00:32:34 2003 Thanks, G. This app is so cool! Actually, the author has been after me since yesterday, and the only way i found to get rid of him was to upload a new version :-P BTW, take care that some part of this software are in PLF now. If any cross-compilation guru knows how to build windows dll on Linux, they could be moved back to contribs, as they are GPL. -- Guillaume Rousse Any line, wire, or cloth cut to length will be too short -- Dimensions General Shefields Corollaries n°1
[Cooker] net-profile doesn't change hostname
The network service doesn't take care of hostname, so just running service network restart won't change it even if it need to be changed in new profile. The attached patch doest take care of this. BTW, is there any reason to use eval `cat $file` instead of more commons . $file or source $file? -- Guillaume Rousse Why do 'tugboats push their barges? -- Why Why Why n°26 --- /sbin/set-netprofile 2003-09-08 18:30:22.0 +0200 +++ set-netprofile 2003-09-11 20:50:50.0 +0200 @@ -68,6 +68,8 @@ done /etc/init.d/network restart + . /etc/sysconfig/network + hostname $HOSTNAME for s in $(start $OLD $NEW); do /etc/init.d/$s start
Re: [Cooker] Re: Re: [Contrib-Rpm] imp3-3.2.2-1mdk
Ainsi parlait David Walser : Guillaume Rousse wrote: Ainsi parlait David Walser : Guillaume Rousse wrote: [Contrib-RPM] Name: imp3 Relocations: (not relocateable) Version : 3.2.2 Vendor: MandrakeSoft Release : 1mdk Build Date: Mon 08 Sep 2003 10:46:41 PM CEST -=-=-=- - drop ugly configuration scripts, perl rulez Well that's fine, but part of the point before was, the script was included with the package, so if the user changed their hostname later on, they could run the script again and have their config be right again. Do the same with your perl script. All the script was doing was to replace localhost.localdomain by the correct values in a config file, using two scripts, a shell one and a awk one. It tooks me almost 10 minutes to understand it, whereas a single one-liner in perl could do in one single place. Moreover, i don't see the point of adding aditional code, meaning potential security and maintainance issues, for such a trivial task, especially for what is supposed to be a server. Either you write a complete configuration wizard, and install it elsewhere in the filesystem, either you just takes cares of it in post-install configuration step. Security and Maintainance issues? That's absolutely ridiculous. Wasting ten minutes to figure that something included was quite useless is an issue. I hate this. Yes your line of code takes care of it in the post-install configuration step, but let me re-iterate, what if the user changes their hostname *after the package is installed.* They will need to run that line of code again. Install it as a script like was done before. Someone changing hist hostname will have more than imp to reconfigure IMHO. Are you proposing we add two script to each of those applications just for this ? Or that we add x others scripts to help clueless user to change x others settings in imp configuration ? My point remain the same: either you provide a full configuration wizard, you document it as mdk-specific, and you install it elsewhere as under the webroot. Or you just do minor post-install configuration. BTW, this is a server, not a joe-user multimedia application. Admins are supposed to be able to configure what they install. -- The only perfect science is hindsight -- Murphy's Laws on Technology n°14
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Chmouel Boudjnah : Guillaume Rousse [EMAIL PROTECTED] writes: Here are two patches for bash completion. just trying would be awesome to update the zsh completion as well /just trying Or at least just make it work. Current zsh package tries to emulate bash-completion, and fails miserabily: [EMAIL PROTECTED] guillaume]$ zsh /etc/bash_completion:53: command not found: shopt /etc/bash_completion:59: command not found: complete /etc/bash_completion:65: command not found: complete /etc/bash_completion:66: command not found: complete /etc/bash_completion:67: command not found: complete ... -- The problem with taking the easy way out is that the enemy has already mined it -- Murphy's Military Laws n°17
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Götz Waschk : Hi, I have a suggestion on how to improve the urpmi completion even more: It would be nice if the completions would also be available for gurpmi, as it's in the normal user's path. It seems to be just an alias for urpmi command, ensuring root privileges and graphical environment. Can i safely use the same completion function as for urpmi ? -- Don't look conspicuous; it draws fire -- Murphy's Military Laws n°28
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Pascal Terjan : Guillaume Rousse wrote: COMPREPLY=( $( compgen -W '-a -p -P -y -s -q -v -h --help \ - --update --media --excludemedia --sortmedia \ - --synthesis --auto --auto-select --fuzzy --src \ - --install-src --clean --noclean --force \ - --allow-nodeps --allow-force --parallel --wget --curl \ - --limit-rate --proxy --proxy-user --bug --env \ - --X --best-output --verify-rpm --test --excludepath' \ - -- $cur ) ) + --update --media --excludemedia --sortmedia --synthesis \ + --auto --auto-select --no-uninstall --keep --split-level \ + --split-length --fuzzy --src --install-src --clean \ + --noclean --force --allow-nodeps --allow-force --parallel \ + --wget --curl --limit-rate --proxy --proxy-user --bug \ + --env --X --best-output --verify-rpm --test --excludepath \ + --excludedocs ' -- $cur ) ) please add --no-verify-rpm which is quite handy for contribs :) Fill a bugreport against urpmi, it isn't in --help output :-) -- The guy that has a bumper sticker that says If you can read this, You're too close always tailgates -- Murphy's Driving Laws n°2
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Götz Waschk : Impressive, it's so fast. It would be nice if this would be included in 9.2. It will, unless someone finds a bug I can't fix in time. BTW, I already know one, but I won't expose it :-) -- Persons disagreeing with your facts are always emotional and employ faulty reasoning -- Murphy's In Laws n°12
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Luca Berra : er, it is bash_completion fault. --- if [ $PS1 ] [ $SHELL = /bin/bash -a -f /etc/bash_completion ]; then --- it should check for $BASH $BASH_VERSION or similar, not $SHELL which is set by login. Good point. Moreover, it should just test if $BASH variable is defined, as even when bash is used as /bin/sh completion is allowed. Please apply attached patch and report any problem. zsh completions should be modified independently from bash. zsh has it own completion metod, and has had it for much longer than bash, i do not use bash, so i will not compare the two. zsh completion primitives are way much better than bash completion primitives. However, it seems there is less active development for zsh completion function, whereas bash-completion project is very active. -- A clean and dry set of overalls is a magnet for mud and rain -- Murphy's Bush Fire Brigade Laws n°16 --- bash-completion.sh~ 2003-09-10 17:25:25.0 +0200 +++ bash-completion.sh 2003-09-10 17:24:46.0 +0200 @@ -1,4 +1,4 @@ -if [ $PS1 ] [ $SHELL = /bin/bash -a -f /etc/bash_completion ]; then +if [ $PS1 ] [ -n $BASH -a -f /etc/bash_completion ]; then # PS1 is set: we're in interactive shell # Source completion code . /etc/bash_completion
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Frederic Crozat : BTW, could you fix bash-completion so : gendiff foobar-x.y | bzip2 -c [TAB] will allow TAB to use ANY (or at least .bz2) file for this redirection. No, i can't. Dealing with redirection has to be managed at bash level directly. However, you still can force standard completion, using meta+/ combination (alt+shift+: on standard keyboard) -- Guillaume Rousse You cannot know both the location of a female and that she is single at any given time. If the location of the female in question is known, see Law #1 -- The Heisenberg Principle of Dating
Re: [Cooker] Enlarge your bash completion experience
Ainsi parlait Chmouel Boudjnah : Guillaume Rousse [EMAIL PROTECTED] writes: However, it seems there is less active development for zsh completion function, whereas bash-completion project is very active. because zsh completion system has been here for years when the bash users start to get so happy with this *new* toy. Seems they managed to push it beyond the older one tough. I guess it is just a question of number of people using bash. -- Guillaume Rousse There is no such thing as an atheist in a foxhole -- Murphy's Military Laws n°119
Re: [Cooker] Re: Enlarge your bash completion experience
Ainsi parlait David Walser : Well if he's gonna do some more hacking/bugfixing on bash-completion, here's another one that doesn't complete currently: mpg321 -@ foo.m3u ogg123 -@ foo.m3u Done. OK, who wants some more :-) ? -- Guillaume Rousse Average temperature increases with the amount of clothing brought -- Murphy's Laws of Camping n°10
Re: [Cooker] Re: Re: [Contrib-Rpm] imp3-3.2.2-1mdk
Ainsi parlait Michael Lothian : Everthough of having a file somewhere that has the hostname is? And then referencing everything to that file so if something changes everything us updated with out reconfiguring anything? hostname is in /etc/sysconfig/network. Technically, what you're proposing is doable. However, i still don't understand why so much attention for this topic. Are you really changing your _servers_ hostname everyday, to need such a toy ? It is clearly a waste of resources for me. Moreover, as sysadmin, I hate black magic changing sensible config stuff behind my back. -- Guillaume Rousse When camping in late fall or winter, your underwear will stay at approximately 35 -- Murphy's Laws of Camping n°23
Re: [Cooker] Re: [Contrib-Rpm] imp3-3.2.2-1mdk
Ainsi parlait David Walser : Guillaume Rousse wrote: [Contrib-RPM] Name: imp3 Relocations: (not relocateable) Version : 3.2.2 Vendor: MandrakeSoft Release : 1mdk Build Date: Mon 08 Sep 2003 10:46:41 PM CEST -=-=-=- Guillaume Rousse [EMAIL PROTECTED] 3.2.2-1mdk - 3.2.2 - speac cleanup You call that a cleanup? Geez, I'd hate to see your room. I admit it is a question of preference, but i really think having standardisation between all horde specs, all macros defined in one place, pretty printed tags, some some comments for anyone reading the spec, and dropping ugly apache 1 post scripts can be called a cleanup - remove useless files from webroot (.dist, doc files, .po) - properly tag localisation files - drop apache1 integration - drop ugly configuration scripts, perl rulez Well that's fine, but part of the point before was, the script was included with the package, so if the user changed their hostname later on, they could run the script again and have their config be right again. Do the same with your perl script. All the script was doing was to replace localhost.localdomain by the correct values in a config file, using two scripts, a shell one and a awk one. It tooks me almost 10 minutes to understand it, whereas a single one-liner in perl could do in one single place. Moreover, i don't see the point of adding aditional code, meaning potential security and maintainance issues, for such a trivial task, especially for what is supposed to be a server. Either you write a complete configuration wizard, and install it elsewhere in the filesystem, either you just takes cares of it in post-install configuration step. -- Any system which depends upon human reliability is unreliable -- SNAFU Equations (JB's Scholastic Laws) n°2
[Cooker] Re: Fix for ImageMagick
Ainsi parlait Giuseppe Ghibò : Just to know, you have tested it? (I'm asking because we are in freeze...). No, i didn't tested it. And contribs are not supposed to be frozen AFAIK. Furthermore previous imp RPM was needing a default configuration to have the address book working without further manual configurations. Could you add further logic in %post for this ? Also It would be nice also to add kronolith to the horde suite... Right. I'll try to do it now i have a standard horde spec template :-) -- Vending machine operators will never have a sense of irony or humor -- Murphy's Laws on Vending Machines n°7
[Cooker] pcmcia-cs 3.2.5
From freshmeat announcement, latest version provide update for orinoco cards, which are very common. Could it be uploaded ? -- Vegetables are healthy. For the ones who can afford it. -- Vegetable-law in Hungary
Re: [Cooker] [HARDWARE] Bewan ADSL PCI support
Ainsi parlait FACORAT Fabrice : The drivers seems to be GPL : http://www.bewan.fr/bewan/utilisateurs/telechargement/pilotes/adsl/linux/A9 04-A888-0.6.1.tar.gz The included firmware is not, and comes from Microelectronics, not Bewan. We're working on a PLF drivers package however. -- The qualities that most attract a woman to a man are usually the same ones she can't stand years later -- Murphy's Laws on Sex n°15
Re: [Cooker] Re: [Contrib-Rpm] dirmngr-0.4.4-5mdk
Ainsi parlait John Keller : Adam Williamson wrote: On Sat, 2003-09-06 at 23:02, Thierry Vignaud wrote: Name: dirmngr Relocations: (not relocateable) Summary : Temporary project to work on GnuPG extensions Description : NewPG is a temporary project to work on GnuPG extensions. It will be merged into the regular GnuPG sources someday. ? From http://www.freshports.org/security/dirmngr/ - Dirmngr is a client for managing and downloading certificate revocation lists (CRLs) for X509 certificates and for downloading the certificates themselves. Dirmngr is usually invoked by gpgsm and in general not used directly. - Maybe this would be good to add before the current description? Done. -- Disk space is always ((Data set size) - 200 MB) -- Ralf's Laws of Observational Astronomy n°3
[Cooker] Enlarge your bash completion experience
Here are two patches for bash completion.
The first one provides those mighty improvments for urpmi (and friends)
completion:
- ligthning-fast package completion using latest urpmi pre-generated package
lists
- avoid querying database when word to complete is clearly a file
- up-to-date option list
- some minor forgotten completion fixed
The second one fixes some problems with service completion, such as trying to
complete past the 3rd word, or completing to non-service files as
/etc/init.d/function.
Please test, and if no serious bug is found, i'll happily commit those patches
to bash-completion package.
--
Guillaume Rousse
When a customer has a large number of specialty locks , thast lock will
require very expensive service and pin kits.
When you buy the kits, you only get to use them one time, or you find that
something you already have can be used instead.
When you don't buy the kits, nothing else that you have will work instead, and
you will have constant problems that would have been avoided by buying the
kits.
By the time you finally buy the kits your lost time will exceed the cost of
the kits by ten fold. Then your customer will go elsewhere.
-- Murphy's Laws of Locksmithing n°21
--- /etc/bash_completion 2003-08-25 01:51:22.0 +0200
+++ bash_completion 2003-09-10 02:25:37.0 +0200
@@ -3604,36 +3604,55 @@
# urpmi media function required by other urpmi functions
#
have urpmq {
-_urpmi_media()
+_urpmi_get_medias()
+{
+ medias=$( awk '/{/ {print $0}' /etc/urpmi/urpmi.cfg 2/dev/null | sed -e 's/ [^ ]\+ {$//' )
+}
+
+_urpmi_medias()
{
- local IFS=$'\t\n'
# return list of available media
- COMPREPLY=( $( awk '/{/ {print $0}' /etc/urpmi/urpmi.cfg 2/dev/null \
- | sed -e s/\( \(file\|removable\|ftp\|http\|rsync\|ssh\):\/\/.*\)\? {$//\
- | grep ^${cur///} ) )
+ local medias
+ # get medias list
+ _urpmi_get_medias
+ # return matching ones
+ COMPREPLY=( $( compgen -W $medias -- ${cur///} ) )
}
_urpmi_packages()
{
- local options
+ # return list of available packages
+ local medias
# find media selection options
for (( i=1; i COMP_CWORD; i++ )); do
- if [[ ${COMP_WORDS[i]} == --@(excludemedia|media) ]]; then
- options=$options ${COMP_WORDS[i]} ${COMP_WORDS[i+1]}
+ if [[ ${COMP_WORDS[i]} == --excludemedia ]]; then
+ _urpmi_get_medias
+ for media in ${COMP_WORDS[i+1]//,/ }; do
+medias=${medias//$media/}
+ done
+ i=$(($i+1))
+ fi
+ if [[ ${COMP_WORDS[i]} == --media ]]; then
+ medias=${COMP_WORDS[i+1]//,/ }
i=$(($i+1))
fi
if [[ ${COMP_WORDS[i]} == --update ]]; then
- options=$options ${COMP_WORDS[i]}
+ COMPREPLY=( $( urpmq --update --list 2/dev/null | grep ^$cur ) )
+ return 0
fi
done
- # return list of available packages
- COMPREPLY=( $( urpmq $options --list 2/dev/null | grep ^$cur ) )
+ # get medias list if still empty
+ [ -z $medias ] _urpmi_get_medias
+
+ # get matching packages
+ for media in $medias; do
+ COMPREPLY=( [EMAIL PROTECTED]:-} $( cat /var/lib/urpmi/names.$media 2/dev/null | grep ^$cur ) )
+ done
}
_urpmi_aliases()
{
- local IFS=$'\t\n'
# return list of available aliases
COMPREPLY=( $( awk -F: '{print $1}' /etc/urpmi/parallel.cfg 2/dev/null | grep ^${cur///} ) )
}
@@ -3651,8 +3670,8 @@
prev=${COMP_WORDS[COMP_CWORD-1]}
case $prev in
- --media)
- _urpmi_media
+ --@(media|excludemedia|sortmedia))
+ _urpmi_medias
return 0
;;
--parallel)
@@ -3664,16 +3683,18 @@
if [[ $cur == -* ]]; then
# return list of available options
COMPREPLY=( $( compgen -W '-a -p -P -y -s -q -v -h --help \
- --update --media --excludemedia --sortmedia \
- --synthesis --auto --auto-select --fuzzy --src \
- --install-src --clean --noclean --force \
- --allow-nodeps --allow-force --parallel --wget --curl \
- --limit-rate --proxy --proxy-user --bug --env \
- --X --best-output --verify-rpm --test --excludepath' \
- -- $cur ) )
+ --update --media --excludemedia --sortmedia --synthesis \
+ --auto --auto-select --no-uninstall --keep --split-level \
+ --split-length --fuzzy --src --install-src --clean \
+ --noclean --force --allow-nodeps --allow-force --parallel \
+ --wget --curl --limit-rate --proxy --proxy-user --bug \
+ --env --X --best-output --verify-rpm --test --excludepath \
+ --excludedocs ' -- $cur ) )
else
- # return rpm files and available packages
- _urpmi_packages
+ # return available packages (unless it is clearly a file) and rpm files
+ if [[ $cur != */* ]]; then
+ _urpmi_packages
+ fi
_filedir rpm
fi
}
@@ -3691,8 +3712,8 @@
prev=${COMP_WORDS[COMP_CWORD-1]}
case $prev in
- --media)
- _urpmi_media
+ --@(media|excludemedia|sortmedia))
+ _urpmi_medias
return 0
;;
--parallel)
@@ -3703,15 +3724,17 @@
if [[ $cur == -* ]]; then
# return list of available options
- COMPREPLY=( $( compgen -W '-h -v -d -u -c -P -R -y -s -i -g \
- -r -f --help --update --media --excludemedia
Re: [Cooker] mdadm-1.3.0-2mdk
Ainsi parlait Luca Berra : hello, i updated mdadm with some fixes from Neil, you might want to update the one in contribs. http://percy.comedia.it/~bluca/cooker/lvm2/ Uploaded mdadm should go into main raidtools should go in contrib Only if initscripts are updated accordingly to support it. I've submitted a patch a long, long time ago for this. -- The only thing you didn't check for a malfunction, will be the source of the problem, but you won't find it until you are called back -- Murphy's Laws of Locksmithing n°3
Re: [Cooker] mdadm-1.3.0-2mdk
Ainsi parlait Oden Eriksson : måndagen den 8 september 2003 11.35 skrev Guillaume Rousse: Ainsi parlait Luca Berra : hello, i updated mdadm with some fixes from Neil, you might want to update the one in contribs. http://percy.comedia.it/~bluca/cooker/lvm2/ Uploaded Huh? I would prefer you ask the maintainer (me) first. As it seems one of the patches applied can possible break usage with 2.4 kernels, that wouldn't be nice. 003MdSuperFix Status: ok Make sure unused superblock descriptor entries aren't failed. This confuses 2.4 kernel code. Sorry. Feel free to remove the patch then if needed. -- Suppressive fire works on everything but the enemy -- Murphy's New Military Laws n°12
Re: [Cooker] Re: perms on /dev/rtc device
Ainsi parlait Juan Quintela : olivier == Olivier Blin [EMAIL PROTECTED] writes: # RTC resolution dev.rtc.max-user-freq = 1024 Could this setting be added in default sysctl.conf ? olivier Thanks, but shouldn't this be the default in default security level ? olivier RTC works fine, but sysctl.conf need to be tweaked. olivier IMHO, the user shouldn't have to do that. Problem is that in a multiuser system, if you allow the value 1024, you can create a DOS if several users use that. I guess most multimedia applications are only usable by local user, not a remote one, which means only one at a time. This should reduce DOS risks, no? What about adding this setting only through mplayer, tvtime and other packages requiring it %post/%postun facilities ? Default value of 64 should be enough except for single-user machines running an _almost_ real time application. And yes, for today machines, mplayer is still real-time like application. Not sure to understand what you mean there. -- Field experience is something you don't get until just after you need it -- Murphy's Bush Fire Brigade Laws n°19
[Cooker] Re: [CHRPM] mailman-2.1.2-7mdk
Ainsi parlait Guillaume Rousse : -=-=-=- Guillaume Rousse [EMAIL PROTECTED] 2.1.2-7mdk - fix my previous fix for postfix, it really needs 'nogroup' for gid - more patching for exotic encoding in attachements - added back missing localised files - constant files under /usr/lib/mailman, variable files under /var/lib/mailman - better mta integration at %post - turned webserver requires to apache, as we use apache specific macros and configuration - additional files as sources, not patches I've changed many things in this release, as previous one was broken anyway... I've tried to adapt %post to automatically adapt to mta used, meaning either postfix or sendmail. However, as i don't have any clue about sendmail (except it uses ugly m4-based configuration), and the README.SENDMAIL is almost empty, I just pushed old script into sendmail-specific part. Could mail gurus check this part ? Other people feedback appreciated too :-) -- If it's worth doing, it's got to be done right now -- Anti- Murphy's Laws n°8
[Cooker] Horde suite
I just rewieved the whole horde suite: - horde2 - imp3 - turba - chora It seems old imp 1 or 2 are no longer around, while there is still old horde-nag 1. Could we nuke it, and turn horde2 to horde and imp3 to imp ? BTW, i found the ADVX macros very useful. Why are those macros in /usr/share/ADVX, and not integrated among standard rpm macros ? I also found most private directories are both protected by the in apache config file, and by a local .htaccess in the directory. Isn't this a bit redundant. Finally, all those sensible files (apache config file, .htaccess files) are owned by apache groupe, with 640 perms, which make rpmlint scream. Time for some new rpmlint exception/check ? -- Guillaume Rousse If the facts do not confirm the theory, they must be disposed of. -- Maier's Law
Re: [Cooker] Re: What happened to am-utils?
Ainsi parlait Luca Berra : no, it is not needed, postfix works wonderfully with mailbox_command = which means that postfix local takes care of deliveries, it even works better than procmail with maildirs the above should be the default and procmail %post and %postun could invoke postconf -e 'mailbox_command = /usr/bin/procmail -Y -a $DOMAIN' and postconf -e 'mailbox_command = ' to add remove procmail support with the package Luca, you should provide ready to use patches for postfix spec, gc would barely reject them i think. -- Guillaume Rousse The pattern you wanted to make again will have one key piece missing -- Murphy's Laws of Sewing n°7
Re: [Cooker] Re: What happened to am-utils?
Ainsi parlait Buchan Milne : The changes to procmail would be more complex, since I assume we don't want procmail to require postfix, so it will need to run postfonf only if postfix is installed (to avoid an error) and run the postconf via triggers if postfix is installed after procmail. What would we avoid procmail to require postfix ? i guess procmail is only useful with a MTA, right ? So a %post taking care of wich MTA is installed (postfix or sendmail) would be enough. %triggers plainly sucks IMHO, as they make maitainance a nightmare. BTW, i'd happily drop sendmail in contrib, and have better postfix integration with mail-related tools. Mailman %post script are a complete mess. -- Guillaume Rousse When reading the instructions of a pump-activated water filter, hour should be substituted for minute when reading the average quarts filtered per minute -- Murphy's Laws of Camping n°17
Re: [Cooker] etc-update
Ainsi parlait Luca Berra : etc-update is a nice idea, but the script is poorly implemented. i uploaded a fixed version at http://percy.comedia.it/~bluca/cooker/misc could it be uploaded in contribs Done. You really should ask a contrib account to lenny :-) -- Guillaume Rousse If it isn't bad for you, it's too expensive -- Murphy's Food Laws n°2
Re: [Cooker] perms on /dev/rtc device
Ainsi parlait Olivier Blin : What perms need to be tweaked to allow this iotcl ? Not a perm problem. Just add the following in your /etc/sysctl.conf: # RTC resolution dev.rtc.max-user-freq = 1024 Could this setting be added in default sysctl.conf ? -- Guillaume Rousse Scattered light never gets into your setup where it is possible -- Ralf's Laws of Observational Astronomy n°8
Re: [Cooker] packager email?
Ainsi parlait Oden Eriksson : Hi. Since when did this change? W: libdspam3-devel invalid-packager Oden Eriksson [EMAIL PROTECTED] The packager email must finish with @mandrakesoft.com or must be [EMAIL PROTECTED] rpmlint complains about this, what should I do? Just ignore it. You can safely add thos two lines in your ~/.rpmlintrc: setOption(Packager, .*) addFilter(E: .* no-signature) -- Guillaume Rousse Does a fish get cramps after eating? -- Why Why Why n°2
Re: [Cooker] perms on /dev/rtc device
Ainsi parlait Han Boetes : Laurent Culioli [EMAIL PROTECTED] wrote: Le Samedi 6 Septembre 2003 12:14, Frederic Crozat a écrit : Yeah, I saw that but I don't think it is related to /dev/rtc.. tvtime is probably trying to nice itself to real-time, which is only permitted for root. Anyway, it still works great and latest version has now a menu.. I think we will soon be able to replace xawtv with tvtime as the default TV app :) in the faq ( http://tvtime.sourceforge.net/help.html#performance ) , there is somes tips to have better performance , but for maximum performance you need to run as root ( there is the same problem with cdrecord if you need to have high priority and buffer on cdwriter device ) , what do you tink to segid video tvtime like cdrecord ? like this : root.video rws r-s r-x ? That's too much. You don't give an app root permissions when it needs real time priority, you give it real time priority. IE make a wrapper. Can you explain it a bit more ? AFAIK, there is privilege separation in default kernel allowing to just give some root privileges, and not others. Moreover, tvtime seems to be designed to drop root privileges once real time priority acquired, see the URL above. -- Guillaume Rousse You can't run a barn without baling twine -- Murphy's Horse Laws n°15
[Cooker] xconfig still broken
* sam sep 06 2003 Juan Quintela [EMAIL PROTECTED] 2.4.22-5mdk - make xconfig works again (lufs is fixed). [EMAIL PROTECTED] linux]# make xconfig ... 3rdparty/lufs/Config.in: 2: unknown command As this error occurs very frequently, could it be tested automatically in any way in the package build sequence ? -- Guillaume Rousse The grocery store always gets a fresh shipment immediately after you buy your food for the week -- Murphy's Food Laws n°9
Re: [Cooker] missing package in kdenetwork
Ainsi parlait Olivier Blin :
For localized files, use this :
%lang(de) %{_datadir}/locale/de/LC_MESSAGES/kwifimanager.mo
%{_prefix}/share/locale/fr/LC_MESSAGES/kwifimanager.mo
%lang(fr) %{_datadir}/locale/fr/LC_MESSAGES/kwifimanager.mo
%{_prefix}/share/locale/ja/LC_MESSAGES/kwifimanager.mo
%{_prefix}/share/locale/it/LC_MESSAGES/kwifimanager.mo
and so on ...
No, use %find_lang macro, see
http://www.linux-mandrake.com/en/howtos/mdk-rpm/more-macros.html
--
Guillaume Rousse
We don't know one millionth of one percent about anything
-- Murphy's Laws on Technology n°32
Re: [Cooker] missing package in kdenetwork
Ainsi parlait Buchan Milne : Maybe on the Wiki, we need a checklist, or list of common mistakes, or maybe a list of common tricks/workarounds? DarkRPMSecrets page, including the infamous avoid nivdia requires trick... -- Guillaume Rousse Program complexity grows until it exceeds the capabilities of the programmer who must maintain it. -- Murphy's Laws of Computer Programming n°13
Re: [Cooker] xconfig still broken
Ainsi parlait Thomas Backlund : its a typo... change xtristate... - tristate... I found it by myself. Now i know that Juan uses vim :-) -- Guillaume Rousse Badness comes in waves -- SNAFU Equations (JB's Scholastic Laws) n°6
Re: [Cooker] missing package in kdenetwork
Ainsi parlait Olivier Blin : On Sat, 6 Sep 2003 14:09:11 +0200 (SAST) Buchan Milne [EMAIL PROTECTED] wrote: BTW, a lot of these issues are covered in the Wiki, if some things are not covered, or need better explaining, let us know, or make the mods to the Wiki. I'm afraid almost nothing of this is covered in the RpmHowTo Wiki :) What needs to be added : - buildrequires issues - menu system - %lang and %find_lang macros - implicit dependencies - libtool nightmare - %mklibname - rpath issues, if gwenole could enlighten us poor mortals that don't read gcc sources everyday on this topic -- Guillaume Rousse When you have secured the area, make sure the enemy knows it too -- Murphy's Military Laws n°110
Re: [Cooker] What is the difference between kernels?
Ainsi parlait Thomas Backlund : From: Jure Repinc [EMAIL PROTECTED] There are a lot of kernels available to install. I have 11 uf I do urpmi kernel: #Theese ar MDK kernels (in main/cooker, and will be on the ISO:s): 1- kernel-2.4.22.5mdk-1-1mdk.i586 - i586optimized, single processor, no highmem = less that 1GB Ram (or actually about 880MB) 2- kernel-enterprise-2.4.22.5mdk-1-1mdk.i586 - i686 optimized, smp support, highmem support = 4GB 3- kernel-i686-up-4GB-2.4.22.5mdk-1-1mdk.i586 - i686 optimized, single processor, highmem =4GB 4- kernel-p3-smp-64GB-2.4.22.5mdk-1-1mdk.i586 - P3 optimized, smp support, highmem =64GB 5- kernel-secure-2.4.22.5mdk-1-1mdk.i586 - i586 optimised, secured with grsec, smp support, no highmem 6- kernel-smp-2.4.22.5mdk-1-1mdk.i586 - i586 optimised, smp support, no highmem What's the interest of providing so much different builds of the same kernel ? Someone with real needs for optimisations such as i686, smp or highmem support is likely to rebuild it anyway. -- Program results should always be reproducible -- Educational Considerations n°5
Re: OT: Communism and authoritarianism (was Re: [Cooker] Re: [CHRPM] rxvt-2.7.10-3mdk)
Ainsi parlait Lyvim Xaphir : Which is splitting hairs and diverting from the main point, which happens to be that the end result of Communism ideology is authoritarianism, wether the members be skinheads, fascists, or authoritarians. Communism is a failed practice, has never worked, has been given ample opportunity to prove itself and has failed, and is basically a stupid ideology. Anybody who endorses communism is an idiot. For that matter anybody that endorces socialism is an idiot. And anybody with such a total lack of political culture should rather shut his mouth than display his ignorance so loudly. -- Guillaume Rousse You never find a lost article until you replace it -- Sad Truths of Life n°8
Re: [Cooker] perms on /dev/rtc device
Ainsi parlait Frederic Crozat : On Mon, 25 Aug 2003 12:14:34 +0200, Jan Ciger wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Frederic Crozat wrote: | console privilege != console user... | console privilege means this user is connected physically on the | system, either by using a text console or a graphical console | (gdm/kdm). Yes, that was what I meant (via pam_console). Is it necessary to limit read access to /dev/rtc to just locally logged in users ? I'm not sure if it is a security risk or not.. Let's first try the safest solution :) mplayer works perfeclty, but there is still a problem with tvtime (from contrib): Can't get realtime priority for better performance, need root access. -- Guillaume Rousse If the vending machine actually has what you want, it will cost more than the amount of change that you have -- Murphy's Laws on Vending Machines n°3
Re: [Cooker] mplayer 0.96-6 bulidrequires libmp3lame-devel
Ainsi parlait Luca Berra : however i cannot find such package, what about requiring liblame-devel? I'm the one to blame here, i changed the name of this package. However, liblame-devel still provides libmp3lame-devel. -- The more complex the idea or technology, the more simple- minded the opposition -- Murphy's In Laws n°16
[Cooker] localisation files for webapps
Several web applications, including all horde suite and mailman, include localisation files under their own directory in /var/www/html. Is there a way to use %find_lang macro so as to look for files elsewhere as under /usr/share/locale ? For my own culture, what's the difference between .mo files included under 'locale' directory, and .po files included under 'po' directory ? -- Why is it that when you're driving and looking for an address, you turn down the volume on the radio? -- Why Why Why n°42
Re: [Cooker] Re: [CHRPM] rxvt-2.7.10-3mdk
Ainsi parlait David Baudens : I can understand that most of people using Cooker need a terminal. But you also need to understand that we don't develop a product only for Cooker users. Most of our users simply don't need a terminal. And why not let newbies choose by themselves ? You'd better provide documentation for educating them, instead of deciding for them. BTW, we are not discussing if newbies need or don't need a terminal, we're constesting the fact that one application should be an exception to general mdk policy, and miss a menu entry just because it could eventually confuse some user. Otherwise we could as well remove menu entry for all browsers except konqueror, for instance... -- the more an item of equipment costs, the farther you have to send it away to be repaired -- Murphy's Bush Fire Brigade Laws n°18
[Cooker] standard macros for webapps
I'm currently upgrading chora, and i have some corrections to do to bugzilla
and mailman package. In order to make maintaince easier, it would be better
to define some standard macro for commonly used directories.
My current proposal is just to make packagers agree on those macros for now,
and enforce them manually in spec file, and later turn them in standard rpm
macros.
The following ones are generic:
%define webroot /var/www
%define webconf %{_sysconfdir}/httpd/conf.d
This one is application-specific:
%define approot %{webroot}/html/%{name}
Some package currently use %prefix for the last one, but i'm not sure if it is
very wise to play with this macro, as it has a specific meaning for rpm. I
could be wrong however.
WDYT ?
--
The degree of technical competence is inversely proportional to the level of
management
-- Murphy's Laws on Technology n°26
[Cooker] kmail-aegypten-plugins
This package is just a a virtual package, taking care of proper dependencies needed for the plugins. It should be moved to kdenetwork-kmail-aegypten-plugins, or maybe just kdenetwork-kmail-plugins, as kmail is actually in kdenetwork-kmail package. Moreover, it would move that really useful feature into main instead of contrib. Many dependencies are redundant: -kdenetwork (which should now be kdenetwork-kmail) already requires kdebase -libgpgme6 already requires gnupg -newpg already requires libgcrypt and libksba -dirmngr already requires libgcrypt and libksba also -cryptplug already requires libgpgme I guess just requiring the following should be ok: -kdenetwork-kmail -cryptplug -dirmngr -newpg -pinentry -- If you take more than your fair share of objectives, you will get more than your fair share of objectives to take -- Murphy's Military Laws n°45
