Discussion:
[gui-dev] new problem of access rights in /cvs/lib
Philippe Verdy
2004-12-04 00:46:33 UTC
Permalink
Note that since today, I'm unable to even get read access to the /cvs/lib
folder, because my account on limewire.org (verdyp, uid=511, group=cvs-lib,
gid=107) is denied all accesses to /cvs (unlike the "guest" user account
which is a member of all groups...)

Something was garbled in access rights...

Normally the "guest" user should be in its own group, where it will only
have read accesses to the various folders. To do that, the cvs folders need
to be effectively owned by the module's group owner, like it is now (and
effectively kept consistent with the "g+s" group-sticky bit as it is now).
But users that are in other groups than the module's owner group should
still be given read access to files, and read/exec access to directories so
that they can be traversed.

On limewire.org, it seems that cvs folders give NO access right at all to
any user outside the file or directory owner or users in the group of the
file or directory.

Normally, there's no reason to forbid read access on these files to
"others".

The "guest" user should not need to be part of the "cvs" or "cvs-lib" or
"cvs-daap" groups. Guest just needs to be in its own default "crapuser"
group: because this group will not own any file or directory (gid), and
because "guest" will also own no file or directory (uid), he will inherit
from access rights from others (so he will have only read access on these
CVS stored files, or read/exec access to these CVS stored folders).

All the limewire team members should already be in the default group
"cvs-private", and then be listed in the "cvs" group to get full control to
CVS, but they may need to be listed also (in /etc/group) in the cvs-lib and
cvs-daap groups.

LimeWire can still, if it wishes so, keep an exclusive list of files visible
only by Limewire crew (sberlin, gbildson, jschmidt, zlatinb, mformel,
kcatillaz, tjones) by setting the container folder of their files to be
owned by someone in the "cvs-private" group, and by listing these users
within the "cvs-private" group (in /etc/group). These protected files and
folders that no one else must see should be given no access right for
"others", but only group-level access rights (read or read-write or
read-exec or read-write-exec).


Consequence: I can't update anything from the lib module. I need to logon as
guest to get at least the read access (cvs update). There's no way to change
anything (cvs add, cvs commit) when logged as "verdyp" or as "guest".

Isn't there someone trained with such basic Unix administration skills at
LimeWire?


Correction to do: restore the read access for others (chmod o+r) on files,
and read-exec access for others on directories. These have incorrectly been
set to zero:

$ cd /spare/cvs
$ chmod -R o+r .
$ find . -type d -exec chmod o+rx {} \;

Then make sure that guest is not a member of all groups that have already
given write accesses to CVS folders:

$ id guest

(normally the group of the "guest" user, uid=502, should be "crapuser",
gid=506; he will inherit of read-only access rights given to 'others' for
files and directories)
Eugene Romanenko
2004-12-04 16:26:01 UTC
Permalink
Hello,

Well, finally completed long-time-ago-promised russian translation.
--
Bye,
Eugene. http://eros2.by.ru
Philippe Verdy
2004-12-04 20:45:35 UTC
Permalink
Hello Eugene,

I did not know that this famous French first name (in French we write it
with an accent as "Eugène", but it's not very common now) was also in use
today in Russia.

Thanks a lot for your contribution, whose completion level at 91.74% puts it
in the top list of translated languages (note that the completion level
ignores the number of translated resources for the optional last part of the
file, used for advanced statistics).

Note also that the meta-data fields in the search panel are also
translatable

Once LimeWire will have reopened the repository, your excellent submission
will be stored, and made available for the next LimeWire beta or release.

Philippe.

----- Original Message -----
From: "Eugene Romanenko" <***@ecomstation.ru>
To: <***@gui.limewire.org>
Sent: Saturday, December 04, 2004 5:26 PM
Subject: [gui-dev] MessagesBundle, russian translation
Post by Eugene Romanenko
Hello,
Well, finally completed long-time-ago-promised russian translation.
--
Bye,
Eugene. http://eros2.by.ru
--------------------------------------------------------------------------------
Post by Eugene Romanenko
_______________________________________________
gui-dev mailing list
http://www.limewire.org/mailman/listinfo/gui-dev
Eugene Romanenko
2004-12-05 10:13:23 UTC
Permalink
Post by Philippe Verdy
I did not know that this famous French first name (in French we write it
with an accent as "Eugène", but it's not very common now) was also in
use today in Russia.
It's common Russian name, but in Russian it pronounced as "Evgenii"
where accent same as in French and last "i" not real "i" but Russian
"short i" which is consonant sound. So, because this name cannot be
written directly in latin, I use English version of this name.
Post by Philippe Verdy
Thanks a lot for your contribution, whose completion level at 91.74%
puts it in the top list of translated languages (note that the
completion level ignores the number of translated resources for the
optional last part of the file, used for advanced statistics).
I hope, translation will help to increase LimeWire popularity in Russia
(or exUSSR). I'm not expect increase of Pro sells, Russia traditionally
not pay for software :), but I hope it will increase Gnutella userbase,
which also good.
Post by Philippe Verdy
Note also that the meta-data fields in the search panel are also
translatable
Yes, I'll translate it later.

And question. I not translated OPTIONS_FORCE_IP strings because they
looks little bit incorrect. (Or I understand it incorrectly.) As I
understand, this option forces EXTERNAL IP to be visible as host IP, but
OPTIONS_FORCE_IP_TITLE declared as "Force Local IP". May be this should
be "Force External IP"?

Correct me, if I think wrong, this option should be used when firewall
defines a portmap from external interface to local LimeWire host? If so,
OPTIONS_FORCE_IP_LABEL should say about portmap need to be defined on
firewall.
--
Bye,
Eugene. http://eros2.by.ru
Philippe Verdy
2004-12-05 18:04:35 UTC
Permalink
Hi Evgenii,

My access to the LimeWire repository has been repaired. I have committed
your translation.
It is already available through the "Russian" link on the translate page.

(Note: the translate page at http://www.limewire.org/translate.shtml is
generated from a script, which is run from time to time, but it has still
not been refreshed. I can run this script, but I can't store its result on
the website.)

Thanks,
Philippe.

Note: don't change the "force local IP" option for now. The English option
is correct because in fact it is related to the setting of the local
end-point that your running host is listening. There's a separate setting
for the public end-point accessible from the internet, in the
Advanced->Firewall settings.

Trust the English message as it is, notably when the message is there since
long. If it makes sense of modifying an English message, the LimeWire team
will do it properly in time with its developments, but you can still discuss
about these messages on the LimeWire development list or on support forums.

At the origin, the local IP could be set in addition to the local port. Now
only thr local port needs to be set, but if you want to have a convenient
translation, translate it as "force local address" (the address designates
both the IP addres of a host interface and a demultiplexing port number on
that interface, associated to the service running on that host and listening
for this interface).

There has been several changes in these options (one of the most complex to
manage in LimeWire), notably because now LimeWire can work more
transparently through proxies, and NAT-routers. In most cases now, such
advanced setting is rarely needed, and LimeWire prefers determining the
correct settings itself, using tests through echoing hosts on the network,
and autoadaptive behavior to change this setting on the fly (notably when
the internet connection uses dynamic and temporarily assigned addresses, or
when the user can't figure out precisely how its external gateway or router
or FAI is configured, or has no easy and understandable way of figuring out
himself those options).
Sam Berlin
2004-12-05 18:17:21 UTC
Permalink
The website is now updated with the more recent translate page. We'll
likely add something to automatically update it every night (as well as
synch the resource bundles of other languages).

The Russian translation has jumped from 1.10% translated to 91.74%
translated. Thank you Eugene and Philippe!

Thanks,
Sam
-----Original Message-----
Sent: Sunday, December 05, 2004 1:05 PM
Subject: Re: [gui-dev] MessagesBundle, russian translation
Hi Evgenii,
My access to the LimeWire repository has been repaired. I have committed
your translation.
It is already available through the "Russian" link on the translate page.
(Note: the translate page at http://www.limewire.org/translate.shtml is
generated from a script, which is run from time to time, but it has still
not been refreshed. I can run this script, but I can't store its result on
the website.)
Thanks,
Philippe.
Note: don't change the "force local IP" option for now. The English option
is correct because in fact it is related to the setting of the local
end-point that your running host is listening. There's a separate setting
for the public end-point accessible from the internet, in the
Advanced->Firewall settings.
Trust the English message as it is, notably when the message is there since
long. If it makes sense of modifying an English message, the LimeWire team
will do it properly in time with its developments, but you can still discuss
about these messages on the LimeWire development list or on support forums.
At the origin, the local IP could be set in addition to the local port. Now
only thr local port needs to be set, but if you want to have a convenient
translation, translate it as "force local address" (the address designates
both the IP addres of a host interface and a demultiplexing port number on
that interface, associated to the service running on that host and listening
for this interface).
There has been several changes in these options (one of the most complex to
manage in LimeWire), notably because now LimeWire can work more
transparently through proxies, and NAT-routers. In most cases now, such
advanced setting is rarely needed, and LimeWire prefers determining the
correct settings itself, using tests through echoing hosts on the network,
and autoadaptive behavior to change this setting on the fly (notably when
the internet connection uses dynamic and temporarily assigned addresses, or
when the user can't figure out precisely how its external gateway or router
or FAI is configured, or has no easy and understandable way of figuring out
himself those options).
_______________________________________________
gui-dev mailing list
http://www.limewire.org/mailman/listinfo/gui-dev
Sam Berlin
2004-12-06 21:24:19 UTC
Permalink
FYI -- Every night at around 10pm EST the translate page will now be
updated and the non-English message bundles will be synchronized to
match the English one.

Thanks,
Sam
Post by Sam Berlin
The website is now updated with the more recent translate page. We'll
likely add something to automatically update it every night (as well as
synch the resource bundles of other languages).
The Russian translation has jumped from 1.10% translated to 91.74%
translated. Thank you Eugene and Philippe!
Thanks,
Sam
-----Original Message-----
Sent: Sunday, December 05, 2004 1:05 PM
Subject: Re: [gui-dev] MessagesBundle, russian translation
Hi Evgenii,
My access to the LimeWire repository has been repaired. I have committed
your translation.
It is already available through the "Russian" link on the translate page.
(Note: the translate page at http://www.limewire.org/translate.shtml is
generated from a script, which is run from time to time, but it has still
not been refreshed. I can run this script, but I can't store its result on
the website.)
Thanks,
Philippe.
Note: don't change the "force local IP" option for now. The English option
is correct because in fact it is related to the setting of the local
end-point that your running host is listening. There's a separate setting
for the public end-point accessible from the internet, in the
Advanced->Firewall settings.
Trust the English message as it is, notably when the message is there since
long. If it makes sense of modifying an English message, the LimeWire team
will do it properly in time with its developments, but you can still discuss
about these messages on the LimeWire development list or on support forums.
At the origin, the local IP could be set in addition to the local
port.
Now
only thr local port needs to be set, but if you want to have a convenient
translation, translate it as "force local address" (the address designates
both the IP addres of a host interface and a demultiplexing port number on
that interface, associated to the service running on that host and listening
for this interface).
There has been several changes in these options (one of the most
complex
to
manage in LimeWire), notably because now LimeWire can work more
transparently through proxies, and NAT-routers. In most cases now, such
advanced setting is rarely needed, and LimeWire prefers determining the
correct settings itself, using tests through echoing hosts on the network,
and autoadaptive behavior to change this setting on the fly (notably when
the internet connection uses dynamic and temporarily assigned
addresses,
or
when the user can't figure out precisely how its external gateway or router
or FAI is configured, or has no easy and understandable way of
figuring
out
himself those options).
_______________________________________________
gui-dev mailing list
http://www.limewire.org/mailman/listinfo/gui-dev
_______________________________________________
gui-dev mailing list
http://www.limewire.org/mailman/listinfo/gui-dev
Sam Berlin
2004-12-05 00:44:47 UTC
Permalink
I'll fix this up. Thought I had tested all the various accounts for
checking out and updating, but apparently I missed some. Sorry for the
troubles, Philippe.

Thanks,
Sam
-----Original Message-----
Sent: Friday, December 03, 2004 7:47 PM
Subject: [gui-dev] new problem of access rights in /cvs/lib
Note that since today, I'm unable to even get read access to the /cvs/lib
folder, because my account on limewire.org (verdyp, uid=511, group=cvs-
lib,
gid=107) is denied all accesses to /cvs (unlike the "guest" user account
which is a member of all groups...)
Something was garbled in access rights...
Normally the "guest" user should be in its own group, where it will only
have read accesses to the various folders. To do that, the cvs folders need
to be effectively owned by the module's group owner, like it is now (and
effectively kept consistent with the "g+s" group-sticky bit as it is now).
But users that are in other groups than the module's owner group should
still be given read access to files, and read/exec access to directories so
that they can be traversed.
On limewire.org, it seems that cvs folders give NO access right at all to
any user outside the file or directory owner or users in the group of the
file or directory.
Normally, there's no reason to forbid read access on these files to
"others".
The "guest" user should not need to be part of the "cvs" or "cvs-lib" or
"cvs-daap" groups. Guest just needs to be in its own default "crapuser"
group: because this group will not own any file or directory (gid), and
because "guest" will also own no file or directory (uid), he will inherit
from access rights from others (so he will have only read access on these
CVS stored files, or read/exec access to these CVS stored folders).
All the limewire team members should already be in the default group
"cvs-private", and then be listed in the "cvs" group to get full control to
CVS, but they may need to be listed also (in /etc/group) in the cvs-lib and
cvs-daap groups.
LimeWire can still, if it wishes so, keep an exclusive list of files visible
only by Limewire crew (sberlin, gbildson, jschmidt, zlatinb, mformel,
kcatillaz, tjones) by setting the container folder of their files to be
owned by someone in the "cvs-private" group, and by listing these users
within the "cvs-private" group (in /etc/group). These protected files and
folders that no one else must see should be given no access right for
"others", but only group-level access rights (read or read-write or
read-exec or read-write-exec).
Consequence: I can't update anything from the lib module. I need to logon as
guest to get at least the read access (cvs update). There's no way to change
anything (cvs add, cvs commit) when logged as "verdyp" or as "guest".
Isn't there someone trained with such basic Unix administration skills at
LimeWire?
Correction to do: restore the read access for others (chmod o+r) on files,
and read-exec access for others on directories. These have incorrectly been
$ cd /spare/cvs
$ chmod -R o+r .
$ find . -type d -exec chmod o+rx {} \;
Then make sure that guest is not a member of all groups that have already
$ id guest
(normally the group of the "guest" user, uid=502, should be "crapuser",
gid=506; he will inherit of read-only access rights given to 'others' for
files and directories)
_______________________________________________
gui-dev mailing list
http://www.limewire.org/mailman/listinfo/gui-dev
Loading...