On 06/10/12 19:20, Chris Rees wrote: > On 10 June 2012 18:10, O. Hartmann <ohartman_at_zedat.fu-berlin.de> wrote: >> On 06/10/12 17:43, John Merryweather Cooper wrote: >>> On 06/10/12 09:54, Martin Sugioarto wrote: >>>> Am Sun, 10 Jun 2012 11:37:09 +0100 >>>> schrieb Chris Rees<crees_at_FreeBSD.org>: >>>> >>>>> Er... people always test their commits. Sometimes edge cases will >>>>> creep in, such as the libreoffice failure which was due to different >>>>> configurations, but to suggest that the commit wasn't tested is quite >>>>> frankly insulting-- it built on a clean system perfectly well. >>>> Hi, >>>> >>>> I don't mean to insult anyone. As I have already told, I am really >>>> thankful that people invest their precious time into updating the ports >>>> collection. >>>> >>>> Whatever "clean system" means. It is surely not the default case that >>>> someone has got a freshly installed set of ports. >>>> >>>> Among all the default problems with ports, libreoffice[1] adds to the >>>> group of annoyances[2] at the moment. I don't know when I have seen >>>> "portmaster -ad" run through successfully last time. I need more and >>>> more "-x" options to exclude ports which fail to build. >>>> >>>> [1] german/libreoffice and libreoffice fails all the time in >>>> (LOCALIZED_LANG is set to "de"): >>>> >>>> Module 'lingucomponent' delivered successfully. 12 files copied, 2 >>>> files unchanged >>>> >>>> ----------------------------------------------------------------------- >>>> Oh dear - something failed during the build - sorry ! >>>> For more help with debugging build errors, please see the section in: >>>> http://wiki.documentfoundation.org/Development >>>> >>>> internal build errors: >>>> >>>> ERROR: error 65280 occurred while >>>> making >>>> /usr/workdir-ports/usr/ports/editors/libreoffice/work/libreoffice-core-3.5.2.2/vcl/prj >>>> >>>> >>>> it seems that the error is inside 'vcl', please re-run build >>>> inside this module to isolate the error and/or test your fix: >>>> ----------------------------------------------------------------------- >>>> >>>> >>>> Whatever this tries to tell me. I don't get it. This is a completely >>>> useless error message for me. >> >> Not even in german/libreoffice. i try to build the standard version and >> I receive the same error. >> >> I can fix this by doing what the buildsystem suggests, but then I have a >> stop in sfx2 and others and it ends up in some module called tail_XXXX, >> where the build never ends when performing the repair as suggested. I >> had once a box running all the night looping building in this folder. >> >>>> >>>> [2] The default annoyances are for example: >>>> >>>> - After updating perl, php or whatever, it makes sense to enforce >>>> updating the modules that belong to these ports. I've seen 100x the >>>> same message that p5-XML-Parser does not work and know what it means, >>>> but this should be resolved by the port system. I mean, when you >>>> update perl, the perl modules won't work anymore. This is totally >>>> clear and it makes sense to update them first before going on. >> >> I can confirm that. I fixed that for me by "portmaster p5-" in case >> p5-SAX-XXX failed. > > There's an UPDATING message written for that very purpose. And even WITH this message written in /usr/ports/UPDATING and follwoing those instrauctions, I have had the very same problem as for years now with this port. The problem is, if you'd like to do an "automated" or "unattended" update of the ports, you stumble very quickly in such a kind of show stopper. If you do not update on a regular basis, those "problems" develop in very serious problems. By the way, the reason why I update also the ports on a regular basis IS because of 100% sure problems if I wait for weeks or months. > >>>> >>>> - When specifying WITHOUT_X11 the ports should respect this and not try >>>> to pull in the X11 variants of ports. I regularly see some ports >>>> pulling ImageMagick instead of the already installed >>>> ImageMagick-nox11. I still do not fully understand what is going on >>>> with WITHOUT_GNOME, but I'll try to figure it out later. But I am >>>> quite sure that some ports pull in unneeded Gnome dependencies. >>>> >>>> - Ports are being marked as interactive and stop the update process. The >>>> idea behind portmaster was (earlier) to avoid interactive building of >>>> ports and ask all the needed questions, before the builds start. I >>>> mean, earlier, I could get out and enjoy some coffee outdoors, now I >>>> have to sit at the keyboard. This is unacceptable! ;) >> >> "portmaster" does even more damage. Sometimed a port reels in some newly >> updates, a port gets deleted. if on of the to be updated prerquisits >> fail, the port in question isn't there anymore. >> >> "portmaster" fails quite often in "oberwriting" remnant files. If a port >> gets corrupted by accident, like graphics/netpbm, One need to delete all >> binaries manually from /usr/local/bin, otherwise the installation fails. >> >> Somehow I wish to have a "brute force" knob to overwrite everything in a >> brutal way. > > FORCE_PKG_REGISTER. Enabled by default in /etc/make.conf in my configuration. And the problem still persists ...
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:27 UTC