[Printing-architecture] Options in /etc/cups/printers.conf

[Printing-architecture] Options in /etc/cups/printers.conf

[Till Kamppeter]

Hi,

I have a problem with non-PPD options set in /etc/cups/printers.conf.

For filters and cups-browsed I have defined some options foor being
stored in /etc/cups/printers.conf which are not options of CUPS itself.

I set these options via

lpadmin -p <printer> -o <option>-default=<value>

getting a line

Option <option> <value>

in the entry of <printer>. What works is that these options get supplied
with every job, so I can use such options to configure filters or to get
filters into a debug mode. I am using this successfully for example with
the "usb" CUPS backend to find quirk rules for broken USB printers.

Now I also want to use this simply to mark print queues (setting an
option which is ignored by all filters). For example all queues created
by cups-browsed get the cups-browsed boolean option set by doing the IPP
equivalent of:

lpadmin -p <printer> -o cups-browsed-default

resulting in a line

Option cups-browsed true

in /etc/cups/printers.conf. This way I want that if cups-browsed not
closed cleanly (system crash) and so did not remove the queues it
created in the next session finds these queues and treats them
correctly, so that crashes do not lead to ghost queues.

Now it seems that I cannot read out this option setting (without sending
a job and checking error_log), as

lpoptions -p <printer>

does not list the option (as "cups-browsed" or "cups-browsed=true") and

val = cupsGetOption("cups-browsed", dest->num_options, dest->options)

in a C program (with dest being the record of the printer in question)
gives val = NULL as one gets when requesting the value of a non-existing
option.

In CUPS 1.5.x (Ubuntu Precise, 12.04 LTS) this still worked. both
cups-browsed was able to read out the "cups-browsed" marker option from
its queues via the C line shown above and

lpoptions -p <printer>

contains "cups-browsed=true" in its output.

I would consider this a bug in CUPS 1.6.x. Mike can you have a look? Thanks.

   Till

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

Filed as Apple bug:

    <rdar://problem/14401795> cups.org: xxx-default options set from lpadmin are not returned by Get-Printer-Attributes

I was able to reproduce both with lpoptions and ipptool.  I'll let you know when I have a fix...


On Jul 9, 2013, at 5:40 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:

> Hi,
> 
> I have a problem with non-PPD options set in /etc/cups/printers.conf.
> 
> For filters and cups-browsed I have defined some options foor being
> stored in /etc/cups/printers.conf which are not options of CUPS itself.
> 
> I set these options via
> 
> lpadmin -p <printer> -o <option>-default=<value>
> 
> getting a line
> 
> Option <option> <value>
> 
> in the entry of <printer>. What works is that these options get supplied
> with every job, so I can use such options to configure filters or to get
> filters into a debug mode. I am using this successfully for example with
> the "usb" CUPS backend to find quirk rules for broken USB printers.
> 
> Now I also want to use this simply to mark print queues (setting an
> option which is ignored by all filters). For example all queues created
> by cups-browsed get the cups-browsed boolean option set by doing the IPP
> equivalent of:
> 
> lpadmin -p <printer> -o cups-browsed-default
> 
> resulting in a line
> 
> Option cups-browsed true
> 
> in /etc/cups/printers.conf. This way I want that if cups-browsed not
> closed cleanly (system crash) and so did not remove the queues it
> created in the next session finds these queues and treats them
> correctly, so that crashes do not lead to ghost queues.
> 
> Now it seems that I cannot read out this option setting (without sending
> a job and checking error_log), as
> 
> lpoptions -p <printer>
> 
> does not list the option (as "cups-browsed" or "cups-browsed=true") and
> 
> val = cupsGetOption("cups-browsed", dest->num_options, dest->options)
> 
> in a C program (with dest being the record of the printer in question)
> gives val = NULL as one gets when requesting the value of a non-existing
> option.
> 
> In CUPS 1.5.x (Ubuntu Precise, 12.04 LTS) this still worked. both
> cups-browsed was able to read out the "cups-browsed" marker option from
> its queues via the C line shown above and
> 
> lpoptions -p <printer>
> 
> contains "cups-browsed=true" in its output.
> 
> I would consider this a bug in CUPS 1.6.x. Mike can you have a look? Thanks.
> 
>   Till

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

[-- Attachment #1: Type: text/plain, Size: 64 bytes --]

Till,

Here is the patch for 1.7 (1.7rc1 to be released soon):


[-- Attachment #2: rdar14401718.patch --]
[-- Type: application/octet-stream, Size: 1235 bytes --]

Index: cups/usersys.c
===================================================================
--- cups/usersys.c	(revision 11112)
+++ cups/usersys.c	(working copy)
@@ -365,6 +365,8 @@
       else if (!strcmp(options, "version=2.2"))
         cg->server_version = 22;
     }
+    else
+      cg->server_version = 20;
 
     if (cg->server[0] != '/' && (port = strrchr(cg->server, ':')) != NULL &&
         !strchr(port, ']') && isdigit(port[1] & 255))
Index: cups/globals.c
===================================================================
--- cups/globals.c	(revision 11112)
+++ cups/globals.c	(working copy)
@@ -215,7 +215,6 @@
   cg->any_root       = 1;
   cg->expired_certs  = 1;
   cg->expired_root   = 1;
-  cg->server_version = 20;
 
  /*
   * Then set directories as appropriate...
Index: cups/ipp.c
===================================================================
--- cups/ipp.c	(revision 11112)
+++ cups/ipp.c	(working copy)
@@ -2340,6 +2340,9 @@
     * Set default version - usually 2.0...
     */
 
+    if (cg->server_version == 0)
+      _cupsSetDefaults();
+
     temp->request.any.version[0] = cg->server_version / 10;
     temp->request.any.version[1] = cg->server_version % 10;
     temp->use                    = 1;

[-- Attachment #3: Type: text/plain, Size: 3318 bytes --]



However, when I went back to test on 1.6 TOT I was unable to reproduce there. Either I had already fixed the problem for 1.6.3 or it wasn't already broken.  The changes in 1.7 (to filter encoded attributes based on operation code) were responsible for the breakage there, but 1.6 doesn't include that code...


On Jul 10, 2013, at 10:17 AM, Michael Sweet <msweet@apple.com> wrote:

> Filed as Apple bug:
> 
>    <rdar://problem/14401795> cups.org: xxx-default options set from lpadmin are not returned by Get-Printer-Attributes
> 
> I was able to reproduce both with lpoptions and ipptool.  I'll let you know when I have a fix...
> 
> 
> On Jul 9, 2013, at 5:40 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:
> 
>> Hi,
>> 
>> I have a problem with non-PPD options set in /etc/cups/printers.conf.
>> 
>> For filters and cups-browsed I have defined some options foor being
>> stored in /etc/cups/printers.conf which are not options of CUPS itself.
>> 
>> I set these options via
>> 
>> lpadmin -p <printer> -o <option>-default=<value>
>> 
>> getting a line
>> 
>> Option <option> <value>
>> 
>> in the entry of <printer>. What works is that these options get supplied
>> with every job, so I can use such options to configure filters or to get
>> filters into a debug mode. I am using this successfully for example with
>> the "usb" CUPS backend to find quirk rules for broken USB printers.
>> 
>> Now I also want to use this simply to mark print queues (setting an
>> option which is ignored by all filters). For example all queues created
>> by cups-browsed get the cups-browsed boolean option set by doing the IPP
>> equivalent of:
>> 
>> lpadmin -p <printer> -o cups-browsed-default
>> 
>> resulting in a line
>> 
>> Option cups-browsed true
>> 
>> in /etc/cups/printers.conf. This way I want that if cups-browsed not
>> closed cleanly (system crash) and so did not remove the queues it
>> created in the next session finds these queues and treats them
>> correctly, so that crashes do not lead to ghost queues.
>> 
>> Now it seems that I cannot read out this option setting (without sending
>> a job and checking error_log), as
>> 
>> lpoptions -p <printer>
>> 
>> does not list the option (as "cups-browsed" or "cups-browsed=true") and
>> 
>> val = cupsGetOption("cups-browsed", dest->num_options, dest->options)
>> 
>> in a C program (with dest being the record of the printer in question)
>> gives val = NULL as one gets when requesting the value of a non-existing
>> option.
>> 
>> In CUPS 1.5.x (Ubuntu Precise, 12.04 LTS) this still worked. both
>> cups-browsed was able to read out the "cups-browsed" marker option from
>> its queues via the C line shown above and
>> 
>> lpoptions -p <printer>
>> 
>> contains "cups-browsed=true" in its output.
>> 
>> I would consider this a bug in CUPS 1.6.x. Mike can you have a look? Thanks.
>> 
>>  Till
> 
> _________________________________________________________
> Michael Sweet, Senior Printing System Engineer, PWG Chair
> 
> _______________________________________________
> Printing-architecture mailing list
> Printing-architecture@lists.linux-foundation.org
> https://lists.linuxfoundation.org/mailman/listinfo/printing-architecture

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Till Kamppeter]

Thank you for the patch, but the patch you have attached is wrong. The
attached patch is for the IPP version probl;em in client.conf (Apple bug
#14401718, Debian bug #711848).

My problem is Apple bug #14401795.

   Till


On 07/10/2013 04:44 PM, Michael Sweet wrote:
> Till,
> 
> Here is the patch for 1.7 (1.7rc1 to be released soon):
> 
> 
> 
> 
> 
> However, when I went back to test on 1.6 TOT I was unable to reproduce there. Either I had already fixed the problem for 1.6.3 or it wasn't already broken.  The changes in 1.7 (to filter encoded attributes based on operation code) were responsible for the breakage there, but 1.6 doesn't include that code...
> 
> 
> On Jul 10, 2013, at 10:17 AM, Michael Sweet <msweet@apple.com> wrote:
> 
>> Filed as Apple bug:
>>
>>    <rdar://problem/14401795> cups.org: xxx-default options set from lpadmin are not returned by Get-Printer-Attributes
>>
>> I was able to reproduce both with lpoptions and ipptool.  I'll let you know when I have a fix...
>>
>>
>> On Jul 9, 2013, at 5:40 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:
>>
>>> Hi,
>>>
>>> I have a problem with non-PPD options set in /etc/cups/printers.conf.
>>>
>>> For filters and cups-browsed I have defined some options foor being
>>> stored in /etc/cups/printers.conf which are not options of CUPS itself.
>>>
>>> I set these options via
>>>
>>> lpadmin -p <printer> -o <option>-default=<value>
>>>
>>> getting a line
>>>
>>> Option <option> <value>
>>>
>>> in the entry of <printer>. What works is that these options get supplied
>>> with every job, so I can use such options to configure filters or to get
>>> filters into a debug mode. I am using this successfully for example with
>>> the "usb" CUPS backend to find quirk rules for broken USB printers.
>>>
>>> Now I also want to use this simply to mark print queues (setting an
>>> option which is ignored by all filters). For example all queues created
>>> by cups-browsed get the cups-browsed boolean option set by doing the IPP
>>> equivalent of:
>>>
>>> lpadmin -p <printer> -o cups-browsed-default
>>>
>>> resulting in a line
>>>
>>> Option cups-browsed true
>>>
>>> in /etc/cups/printers.conf. This way I want that if cups-browsed not
>>> closed cleanly (system crash) and so did not remove the queues it
>>> created in the next session finds these queues and treats them
>>> correctly, so that crashes do not lead to ghost queues.
>>>
>>> Now it seems that I cannot read out this option setting (without sending
>>> a job and checking error_log), as
>>>
>>> lpoptions -p <printer>
>>>
>>> does not list the option (as "cups-browsed" or "cups-browsed=true") and
>>>
>>> val = cupsGetOption("cups-browsed", dest->num_options, dest->options)
>>>
>>> in a C program (with dest being the record of the printer in question)
>>> gives val = NULL as one gets when requesting the value of a non-existing
>>> option.
>>>
>>> In CUPS 1.5.x (Ubuntu Precise, 12.04 LTS) this still worked. both
>>> cups-browsed was able to read out the "cups-browsed" marker option from
>>> its queues via the C line shown above and
>>>
>>> lpoptions -p <printer>
>>>
>>> contains "cups-browsed=true" in its output.
>>>
>>> I would consider this a bug in CUPS 1.6.x. Mike can you have a look? Thanks.
>>>
>>>  Till
>>
>> _________________________________________________________
>> Michael Sweet, Senior Printing System Engineer, PWG Chair
>>
>> _______________________________________________
>> Printing-architecture mailing list
>> Printing-architecture@lists.linux-foundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/printing-architecture
> 
> _________________________________________________________
> Michael Sweet, Senior Printing System Engineer, PWG Chair
> 

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

[-- Attachment #1: Type: text/plain, Size: 24 bytes --]

Till,

Whoops, sorry:



[-- Attachment #2: rdar14401795.patch --]
[-- Type: application/octet-stream, Size: 406 bytes --]

Index: cups/encode.c
===================================================================
--- cups/encode.c	(revision 11114)
+++ cups/encode.c	(working copy)
@@ -534,7 +534,7 @@
       else
         ops ++;
 
-    if (*ops == IPP_OP_CUPS_NONE)
+    if (*ops == IPP_OP_CUPS_NONE && op != IPP_OP_CUPS_NONE)
     {
       DEBUG_printf(("2cupsEncodeOptions2: Skipping \"%s\".", option->name));
       continue;

[-- Attachment #3: Type: text/plain, Size: 4235 bytes --]



Also, I now have an official GIT mirror of live SVN up at:

    http://www.cups.org/cups.git

I will be posting information on the download page shortly...


On Jul 10, 2013, at 11:28 AM, Till Kamppeter <till.kamppeter@gmail.com> wrote:

> Thank you for the patch, but the patch you have attached is wrong. The
> attached patch is for the IPP version probl;em in client.conf (Apple bug
> #14401718, Debian bug #711848).
> 
> My problem is Apple bug #14401795.
> 
>   Till
> 
> 
> On 07/10/2013 04:44 PM, Michael Sweet wrote:
>> Till,
>> 
>> Here is the patch for 1.7 (1.7rc1 to be released soon):
>> 
>> 
>> 
>> 
>> 
>> However, when I went back to test on 1.6 TOT I was unable to reproduce there. Either I had already fixed the problem for 1.6.3 or it wasn't already broken.  The changes in 1.7 (to filter encoded attributes based on operation code) were responsible for the breakage there, but 1.6 doesn't include that code...
>> 
>> 
>> On Jul 10, 2013, at 10:17 AM, Michael Sweet <msweet@apple.com> wrote:
>> 
>>> Filed as Apple bug:
>>> 
>>>   <rdar://problem/14401795> cups.org: xxx-default options set from lpadmin are not returned by Get-Printer-Attributes
>>> 
>>> I was able to reproduce both with lpoptions and ipptool.  I'll let you know when I have a fix...
>>> 
>>> 
>>> On Jul 9, 2013, at 5:40 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:
>>> 
>>>> Hi,
>>>> 
>>>> I have a problem with non-PPD options set in /etc/cups/printers.conf.
>>>> 
>>>> For filters and cups-browsed I have defined some options foor being
>>>> stored in /etc/cups/printers.conf which are not options of CUPS itself.
>>>> 
>>>> I set these options via
>>>> 
>>>> lpadmin -p <printer> -o <option>-default=<value>
>>>> 
>>>> getting a line
>>>> 
>>>> Option <option> <value>
>>>> 
>>>> in the entry of <printer>. What works is that these options get supplied
>>>> with every job, so I can use such options to configure filters or to get
>>>> filters into a debug mode. I am using this successfully for example with
>>>> the "usb" CUPS backend to find quirk rules for broken USB printers.
>>>> 
>>>> Now I also want to use this simply to mark print queues (setting an
>>>> option which is ignored by all filters). For example all queues created
>>>> by cups-browsed get the cups-browsed boolean option set by doing the IPP
>>>> equivalent of:
>>>> 
>>>> lpadmin -p <printer> -o cups-browsed-default
>>>> 
>>>> resulting in a line
>>>> 
>>>> Option cups-browsed true
>>>> 
>>>> in /etc/cups/printers.conf. This way I want that if cups-browsed not
>>>> closed cleanly (system crash) and so did not remove the queues it
>>>> created in the next session finds these queues and treats them
>>>> correctly, so that crashes do not lead to ghost queues.
>>>> 
>>>> Now it seems that I cannot read out this option setting (without sending
>>>> a job and checking error_log), as
>>>> 
>>>> lpoptions -p <printer>
>>>> 
>>>> does not list the option (as "cups-browsed" or "cups-browsed=true") and
>>>> 
>>>> val = cupsGetOption("cups-browsed", dest->num_options, dest->options)
>>>> 
>>>> in a C program (with dest being the record of the printer in question)
>>>> gives val = NULL as one gets when requesting the value of a non-existing
>>>> option.
>>>> 
>>>> In CUPS 1.5.x (Ubuntu Precise, 12.04 LTS) this still worked. both
>>>> cups-browsed was able to read out the "cups-browsed" marker option from
>>>> its queues via the C line shown above and
>>>> 
>>>> lpoptions -p <printer>
>>>> 
>>>> contains "cups-browsed=true" in its output.
>>>> 
>>>> I would consider this a bug in CUPS 1.6.x. Mike can you have a look? Thanks.
>>>> 
>>>> Till
>>> 
>>> _________________________________________________________
>>> Michael Sweet, Senior Printing System Engineer, PWG Chair
>>> 
>>> _______________________________________________
>>> Printing-architecture mailing list
>>> Printing-architecture@lists.linux-foundation.org
>>> https://lists.linuxfoundation.org/mailman/listinfo/printing-architecture
>> 
>> _________________________________________________________
>> Michael Sweet, Senior Printing System Engineer, PWG Chair
>> 
> 

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Till Kamppeter]

Works perfectly, thank you very much for the quick fix.

   Till


On 07/10/2013 05:47 PM, Michael Sweet wrote:
> Till,
> 
> Whoops, sorry:
> 
> Also, I now have an official GIT mirror of live SVN up at:
> 
>     http://www.cups.org/cups.git
> 
> I will be posting information on the download page shortly...
> 

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Till Kamppeter]

Unfortunately, this patch applies to a completely new concept of CUPS
1.7.x which did not exist in 1.6.x. Can you also supply a fix for CUPS
1.6.x, so that I can do a patch release of CUPS for Ubuntu 13.04? Thanks.

   Till

On 07/10/2013 04:44 PM, Michael Sweet wrote:
> Till,
> 
> Here is the patch for 1.7 (1.7rc1 to be released soon):
> 

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

Till,

I don't know if you saw my followup message or not - I was unable to reproduce on TOT 1.6.  You can test TOT 1.6 from the GIT repository with:

    git clone -b branch-1.6 http://www.cups.org/cups.git cups-1.6


On Jul 10, 2013, at 1:11 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:

> Unfortunately, this patch applies to a completely new concept of CUPS
> 1.7.x which did not exist in 1.6.x. Can you also supply a fix for CUPS
> 1.6.x, so that I can do a patch release of CUPS for Ubuntu 13.04? Thanks.
> 
>   Till
> 
> On 07/10/2013 04:44 PM, Michael Sweet wrote:
>> Till,
>> 
>> Here is the patch for 1.7 (1.7rc1 to be released soon):
>> 

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Till Kamppeter]

Sorry, on both machines where I observed the bug I was already on CUPS
1.7.x. I have downgraded to the official Ubuntu package (1.6.2) on my
Saucy (13.10) machine and there the bug indeed did not show. So the
problem got introduced in 1.7.x and I will apply your patch to my 1.7.x
packages on my PPA.

Thank you for the patch anyway, as now the bug will not appear when
distros switch over to 1.7.x.

   Till

On 07/10/2013 07:30 PM, Michael Sweet wrote:
> Till,
> 
> I don't know if you saw my followup message or not - I was unable to reproduce on TOT 1.6.  You can test TOT 1.6 from the GIT repository with:
> 
>     git clone -b branch-1.6 http://www.cups.org/cups.git cups-1.6
> 
> 
> On Jul 10, 2013, at 1:11 PM, Till Kamppeter <till.kamppeter@gmail.com> wrote:
> 
>> Unfortunately, this patch applies to a completely new concept of CUPS
>> 1.7.x which did not exist in 1.6.x. Can you also supply a fix for CUPS
>> 1.6.x, so that I can do a patch release of CUPS for Ubuntu 13.04? Thanks.
>>
>>   Till
>>
>> On 07/10/2013 04:44 PM, Michael Sweet wrote:
>>> Till,
>>>
>>> Here is the patch for 1.7 (1.7rc1 to be released soon):
>>>
> 
> _________________________________________________________
> Michael Sweet, Senior Printing System Engineer, PWG Chair
> 

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Tim Waugh]

[-- Attachment #1: Type: text/plain, Size: 455 bytes --]

On Wed, 2013-07-10 at 11:47 -0400, Michael Sweet wrote: 
> Also, I now have an official GIT mirror of live SVN up at:
> 
>     http://www.cups.org/cups.git
> 
> I will be posting information on the download page shortly...

This only has history starting from May this year.  Is there any plan to
have a single git repository containing all of the SVN history, like the
one I have at http://twaugh.fedorapeople.org/cups-avahi.git?

Tim.
*/


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 482 bytes --]

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

Tim,

On 2013-07-11, at 4:48 AM, Tim Waugh <twaugh@redhat.com> wrote:

> On Wed, 2013-07-10 at 11:47 -0400, Michael Sweet wrote: 
>> Also, I now have an official GIT mirror of live SVN up at:
>> 
>>    http://www.cups.org/cups.git
>> 
>> I will be posting information on the download page shortly...
> 
> This only has history starting from May this year.  Is there any plan to
> have a single git repository containing all of the SVN history, like the
> one I have at http://twaugh.fedorapeople.org/cups-avahi.git?


Unfortunately, I was unable to coerce GIT-SVN into pulling from a single repository.  The CUPS SVN repository has been split since 2002 (easysw.com and apple.com), and so what I ended up doing is pulling all of the snapshots and releases from easysw.com and the commits from apple.com since late April/early May.

If you know of some GIT-fu that will allow me to push the SVN history from the old repository onto the new one (or start with easysw.com's SVN repo and then overlay and track new changes from the apple.com SVN repo) then I'll be happy to rebuild the GIT mirror.  But going forward everything is setup to mirror changes from the internal apple.com SVN repository to the GIT repository on cups.org...

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Tim Waugh]

[-- Attachment #1: Type: text/plain, Size: 657 bytes --]

On Thu, 2013-07-11 at 09:11 -0400, Michael Sweet wrote:
> Unfortunately, I was unable to coerce GIT-SVN into pulling from a
> single repository. The CUPS SVN repository has been split since 2002
> (easysw.com and apple.com), and so what I ended up doing is pulling
> all of the snapshots and releases from easysw.com and the commits from
> apple.com since late April/early May.

I haven't tried that. I would have probably done two git-svn clones, one
for each SVN repository, then tried to merge the newer into the older...
but the merge point would have to be *exactly* the same and I guess
that's where you had problems with it.

Tim.
*/


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 482 bytes --]

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

Tim,

On Jul 16, 2013, at 10:33 AM, Tim Waugh <twaugh@redhat.com> wrote:
> On Thu, 2013-07-11 at 09:11 -0400, Michael Sweet wrote:
>> Unfortunately, I was unable to coerce GIT-SVN into pulling from a
>> single repository. The CUPS SVN repository has been split since 2002
>> (easysw.com and apple.com), and so what I ended up doing is pulling
>> all of the snapshots and releases from easysw.com and the commits from
>> apple.com since late April/early May.
> 
> I haven't tried that. I would have probably done two git-svn clones, one
> for each SVN repository, then tried to merge the newer into the older...
> but the merge point would have to be *exactly* the same and I guess
> that's where you had problems with it.


Actually, I was able to synchronize the SVN revisions between the two repositories (cups.org had a lot more commits than the internal Apple repo), but the issue is that the directory structure of the internal Apple repo is different than the cups.org repo.

To make matters worse, it doesn't look like GIT-SVN is properly mirroring things with the current GIT repository and I think I need to redo the whole thing again to make GIT happy...

I am going to try doing some SVN hackery to backup and restore the cups.org repository to match the Apple SVN repo structure, in case that allows me to provide a fuller history of things...  And then see if I can provide a stable GIT-SVN mirror...

(If that doesn't work then I'll just have to pull the GIT mirror for now; I can't provide a SVN mirror of the internal SVN repository...)

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Tim Waugh]

[-- Attachment #1: Type: text/plain, Size: 390 bytes --]

On Tue, 2013-07-16 at 10:48 -0400, Michael Sweet wrote:
> To make matters worse, it doesn't look like GIT-SVN is properly
> mirroring things with the current GIT repository and I think I need to
> redo the whole thing again to make GIT happy...

I've found that '--stdlayout' is the magic option to make things work
right, every time I've used git-svn to date. *shrug*

Tim.
*/


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 482 bytes --]

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

Tim,

On 2013-07-16, at 12:27 PM, Tim Waugh <twaugh@redhat.com> wrote:
> On Tue, 2013-07-16 at 10:48 -0400, Michael Sweet wrote:
>> To make matters worse, it doesn't look like GIT-SVN is properly
>> mirroring things with the current GIT repository and I think I need to
>> redo the whole thing again to make GIT happy...
> 
> I've found that '--stdlayout' is the magic option to make things work
> right, every time I've used git-svn to date. *shrug*

That only works if you have a standard layout... :/  The internal Apple repo doesn't have a standard layout due to it having more than just cups.org content, so my next attempt will be to try to convert the old repo into the new layout, convert to GIT, and then try to merge changes from the internal repo...

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[James Cloos]

>>>>> "MS" == Michael Sweet <msweet@apple.com> writes:

MS> If you know of some GIT-fu that will allow me to push the SVN history
MS> from the old repository onto the new one (or start with easysw.com's
MS> SVN repo and then overlay and track new changes from the apple.com SVN
MS> repo) then I'll be happy to rebuild the GIT mirror.

It is probably doable.  Publishing a git repo with the the old data in
parallel with the current repo would help get the surgery planned.

Also, it is possible to update www.cups.org to support git's smart
git/http protocol?  (Cf the man page for git-http-backend.  It runs
as a CGI.)  It would improve clones and pulls, as well as provide
users with progress info akin to that seem when using git:// URIs.

-JimC
--
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: [Printing-architecture] Options in /etc/cups/printers.conf

[Michael Sweet]

[-- Attachment #1: Type: text/plain, Size: 1373 bytes --]

James,

I did end up discovering how to combine the history of both repositories.

Now I am just resolving some issues with the two active SVN branches not updating their corresponding Git branches - I can get one to update but not the other (so trunk is tracking properly but the 1.7 branch isn't...)


On Feb 25, 2014, at 6:00 PM, James Cloos <cloos@jhcloos.com> wrote:

>>>>>> "MS" == Michael Sweet <msweet@apple.com> writes:
> 
> MS> If you know of some GIT-fu that will allow me to push the SVN history
> MS> from the old repository onto the new one (or start with easysw.com's
> MS> SVN repo and then overlay and track new changes from the apple.com SVN
> MS> repo) then I'll be happy to rebuild the GIT mirror.
> 
> It is probably doable.  Publishing a git repo with the the old data in
> parallel with the current repo would help get the surgery planned.
> 
> Also, it is possible to update www.cups.org to support git's smart
> git/http protocol?  (Cf the man page for git-http-backend.  It runs
> as a CGI.)  It would improve clones and pulls, as well as provide
> users with progress info akin to that seem when using git:// URIs.
> 
> -JimC
> --
> James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

_________________________________________________________
Michael Sweet, Senior Printing System Engineer, PWG Chair


[-- Attachment #2: smime.p7s --]
[-- Type: application/pkcs7-signature, Size: 4881 bytes --]