[arch-dev-public] Updating 'man'
Hey guys, I tried contacting Andy directly but got no response, so I am going to run this by everyone else. I want to update the man package with a couple of changes: Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully. Run configure with +fhs so that it thinks man pages are in /usr/share/man *first* Is this acceptable? Thanks, Aaron
Am Freitag, 11. Januar 2008 22:11:37 schrieb Aaron Griffin:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully.
Afaik this is no solution. In fact man pages are even more broken then. :-) -- archlinux.de
On Jan 11, 2008 3:24 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Freitag, 11. Januar 2008 22:11:37 schrieb Aaron Griffin:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully.
Afaik this is no solution. In fact man pages are even more broken then. :-)
It fixes *some* here. For instance, take a look at the pacman 3.1 man page. A few of the quote characters are in utf8, and show up as (?) or something until you remove the -Tlatin1. Either way, removing that should be the proper way to go here. Replacing man with man-db is probably more proper, but I am simply working with what we have at this moment, as I don't have enough gusto to do the complete replacement.
On Jan 11, 2008 3:30 PM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Jan 11, 2008 3:24 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Freitag, 11. Januar 2008 22:11:37 schrieb Aaron Griffin:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully.
Afaik this is no solution. In fact man pages are even more broken then. :-)
It fixes *some* here. For instance, take a look at the pacman 3.1 man page. A few of the quote characters are in utf8, and show up as (?) or something until you remove the -Tlatin1.
Either way, removing that should be the proper way to go here. Replacing man with man-db is probably more proper, but I am simply working with what we have at this moment, as I don't have enough gusto to do the complete replacement.
Whoops, hit "send" too soon. Could you please explain what "more broken" means. At the very least, a few of us (eliott, dan, and myself) removed this line a few days ago and I have heard no report of issues. This may be that we're all en_US people, but it may not. You also need to unset LESSCHARSET for the full effect, but this may not be done if you have the old /etc/profile due the the pacman 3.0 backup issue
Am Freitag, 11. Januar 2008 22:32:28 schrieb Aaron Griffin:
You also need to unset LESSCHARSET for the full effect, but this may not be done if you have the old /etc/profile due the the pacman 3.0 backup issue
Ah, good hint. With unsetting LESSCHARSET everything seems fine. OK, when we release a new filesystem without LESSCHARSET I am fine with this change. :-) -- archlinux.de
On Jan 11, 2008 3:39 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Freitag, 11. Januar 2008 22:32:28 schrieb Aaron Griffin:
You also need to unset LESSCHARSET for the full effect, but this may not be done if you have the old /etc/profile due the the pacman 3.0 backup issue
Ah, good hint. With unsetting LESSCHARSET everything seems fine. OK, when we release a new filesystem without LESSCHARSET I am fine with this change. :-)
Yeah, I have other things for filesystem too. 8) I was going to push a new filesystem and bash last night, but I didn't get a chance. I have filesystem, bash, initscripts, and man updates coming down the pipeline 8)
Also of note: pacman 3.1 doesn't appear to overwrite the /etc/profile file, even if you don't change it, because the package changed hands (profile != profile.bash). When I enabled testing, fetched pacman 3.1, and then updated, I got a profile.pacnew. Moving that over to replace the existing file and logout-login fixed all the man pages that I had chance to look at. pacman, git, and a few others.
eliott wrote:
Also of note: pacman 3.1 doesn't appear to overwrite the /etc/profile file, even if you don't change it, because the package changed hands (profile != profile.bash).
When I enabled testing, fetched pacman 3.1, and then updated, I got a profile.pacnew. Moving that over to replace the existing file and logout-login fixed all the man pages that I had chance to look at. pacman, git, and a few others.
Ah indeed, when pacman installs the new filesystem package, it doesn't have the original md5sum of /etc/profile, because that one is only available in the old bash package, and not in the old filesystem package. So instead if being in the following case : original=X, current=X, new=Y The current file is the same as the original but the new one differs. Since the user did not ever modify the file, and the new one may contain improvements or bugfixes, install the new file. original is actually empty, so we end up in the last case: original=X, current=Y, new=Z All three files are different, so install the new file with a .pacnew extension and warn the user. The user must then manually merge any necessary changes into the original file. But in my opinion, that behavior is good enough.
On Jan 12, 2008 3:58 AM, Xavier <shiningxc@gmail.com> wrote:
eliott wrote:
Also of note: pacman 3.1 doesn't appear to overwrite the /etc/profile file, even if you don't change it, because the package changed hands (profile != profile.bash).
When I enabled testing, fetched pacman 3.1, and then updated, I got a profile.pacnew. Moving that over to replace the existing file and logout-login fixed all the man pages that I had chance to look at. pacman, git, and a few others.
Ah indeed, when pacman installs the new filesystem package, it doesn't have the original md5sum of /etc/profile, because that one is only available in the old bash package, and not in the old filesystem package.
Good catch. I can bump the md5sum but fiddling with the file, if people think that is ideal?
Am Freitag, 11. Januar 2008 22:30:13 schrieb Aaron Griffin:
On Jan 11, 2008 3:24 PM, Pierre Schmitz <pierre@archlinux.de> wrote:
Am Freitag, 11. Januar 2008 22:11:37 schrieb Aaron Griffin:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully.
Afaik this is no solution. In fact man pages are even more broken then. :-)
It fixes *some* here. For instance, take a look at the pacman 3.1 man page. A few of the quote characters are in utf8, and show up as (?) or something until you remove the -Tlatin1.
When I remove -Tlatin1 I got the following result: http://users.archlinux.de/~pierre/tmp/man.png Yes, even english man pages are broken now. With -Tlatin1 only those which are localized are broken.
Either way, removing that should be the proper way to go here. Replacing man with man-db is probably more proper, but I am simply working with what we have at this moment, as I don't have enough gusto to do the complete replacement.
I did not test man-db so far. (The package in [community] does not work) -- archlinux.de
Aaron Griffin wrote:
Hey guys, I tried contacting Andy directly but got no response, so I am going to run this by everyone else.
I want to update the man package with a couple of changes:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully. Run configure with +fhs so that it thinks man pages are in /usr/share/man *first*
Is this acceptable?
Removing -Tlatin1 from the nroff call is apparently required for displaying man pages correctly (eg pacman man page with ´ or localized mplayer man page), so that change looks good to me. And probably this should be done when the new filesystem is moved to core.
On Jan 23, 2008 5:38 AM, Xavier <shiningxc@gmail.com> wrote:
Aaron Griffin wrote:
Hey guys, I tried contacting Andy directly but got no response, so I am going to run this by everyone else.
I want to update the man package with a couple of changes:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully. Run configure with +fhs so that it thinks man pages are in /usr/share/man *first*
Is this acceptable?
Removing -Tlatin1 from the nroff call is apparently required for displaying man pages correctly (eg pacman man page with ´ or localized mplayer man page), so that change looks good to me. And probably this should be done when the new filesystem is moved to core.
This is all unrelated to filesystem moving to core. The filesystem change is regarding the LOCATION of man pages, this is regarding the ENCODING of man pages. They're very different issues. This will be done after the filesystem change as we also need to make goofy tmac changes to groff or something... I dunno what happened to that thread. Dan and I figured out what we needed a while back.
On Jan 23, 2008 10:47 AM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Jan 23, 2008 5:38 AM, Xavier <shiningxc@gmail.com> wrote:
Aaron Griffin wrote:
Hey guys, I tried contacting Andy directly but got no response, so I am going to run this by everyone else.
I want to update the man package with a couple of changes:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully. Run configure with +fhs so that it thinks man pages are in /usr/share/man *first*
Is this acceptable?
Removing -Tlatin1 from the nroff call is apparently required for displaying man pages correctly (eg pacman man page with ´ or localized mplayer man page), so that change looks good to me. And probably this should be done when the new filesystem is moved to core.
This is all unrelated to filesystem moving to core. The filesystem change is regarding the LOCATION of man pages, this is regarding the ENCODING of man pages. They're very different issues.
Um...we removed some MANLOCALE setting or something that was latin1 in /etc/profile...so it is related. That is what is causing breakage too. -Dan
On Jan 23, 2008 10:59 AM, Dan McGee <dpmcgee@gmail.com> wrote:
On Jan 23, 2008 10:47 AM, Aaron Griffin <aaronmgriffin@gmail.com> wrote:
On Jan 23, 2008 5:38 AM, Xavier <shiningxc@gmail.com> wrote:
Aaron Griffin wrote:
Hey guys, I tried contacting Andy directly but got no response, so I am going to run this by everyone else.
I want to update the man package with a couple of changes:
Remove -Tlatin from the nroff call, as nroff will autodetect the proper charset (this will help in displaying utf8 man pages), closing 7477 hopefully. Run configure with +fhs so that it thinks man pages are in /usr/share/man *first*
Is this acceptable?
Removing -Tlatin1 from the nroff call is apparently required for displaying man pages correctly (eg pacman man page with ´ or localized mplayer man page), so that change looks good to me. And probably this should be done when the new filesystem is moved to core.
This is all unrelated to filesystem moving to core. The filesystem change is regarding the LOCATION of man pages, this is regarding the ENCODING of man pages. They're very different issues.
Um...we removed some MANLOCALE setting or something that was latin1 in /etc/profile...so it is related. That is what is causing breakage too.
LESSCHARSET But that only affects less, and not man pages. It's a fine line, but I know at least a few people that don't use less as their MANPAGER
Is there any certain reason why we don't stick with the /etc/man.conf file for all man related configuration? Why do we abuse /etc/profile for man's settings? We make it more complicated for non bash users that way. -Andy
On Jan 24, 2008 1:12 AM, Andreas Radke <a.radke@arcor.de> wrote:
Is there any certain reason why we don't stick with the /etc/man.conf file for all man related configuration? Why do we abuse /etc/profile for man's settings? We make it more complicated for non bash users that way.
What? Isn't that the direction we are heading? We are *getting rid of* things like LESSCHARSET and MANPATH, not adding them. -Dan
On Jan 24, 2008 7:09 AM, Dan McGee <dpmcgee@gmail.com> wrote:
On Jan 24, 2008 1:12 AM, Andreas Radke <a.radke@arcor.de> wrote:
Is there any certain reason why we don't stick with the /etc/man.conf file for all man related configuration? Why do we abuse /etc/profile for man's settings? We make it more complicated for non bash users that way.
What? Isn't that the direction we are heading? We are *getting rid of* things like LESSCHARSET and MANPATH, not adding them.
This is correct. The new /etc/profile explicitly UNSETS the variables that would affect man and the supported pagers. Dan is correct here. We are moving towards full man.conf usage
2008/1/24, Aaron Griffin <aaronmgriffin@gmail.com>:
On Jan 24, 2008 7:09 AM, Dan McGee <dpmcgee@gmail.com> wrote:
On Jan 24, 2008 1:12 AM, Andreas Radke <a.radke@arcor.de> wrote:
Is there any certain reason why we don't stick with the /etc/man.conf file for all man related configuration? Why do we abuse /etc/profile for man's settings? We make it more complicated for non bash users that way.
What? Isn't that the direction we are heading? We are *getting rid of* things like LESSCHARSET and MANPATH, not adding them.
This is correct. The new /etc/profile explicitly UNSETS the variables that would affect man and the supported pagers. Dan is correct here. We are moving towards full man.conf usage
Status on man/bash/filesysem update? Users are starting asking why man doesn't work for them more frequently. -- Roman Kyrylych (Роман Кирилич)
On Jan 26, 2008 2:53 PM, Roman Kyrylych <roman.kyrylych@gmail.com> wrote:
Status on man/bash/filesysem update? Users are starting asking why man doesn't work for them more frequently.
Only Pierre responded to my previous signoff message. I will rebuild all three of these and push them to testing because that may be needed.
participants (7)
-
Aaron Griffin
-
Andreas Radke
-
Dan McGee
-
eliott
-
Pierre Schmitz
-
Roman Kyrylych
-
Xavier