[PATCH 0/1] multipath: libdmmp json fix

[PATCH 0/1] multipath: libdmmp json fix

[Benjamin Marzinski]

This is a patch I got to keep libdmmp working with the latest json-c
release.

Björn Esser (1):
  libdmmp: Add support for upcoming json-c 0.14.0.

 libdmmp/libdmmp_private.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

-- 
2.17.2

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

[PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Benjamin Marzinski]

From: Björn Esser <besser82@fedoraproject.org>

TRUE/FALSE are not defined anymore.  1 and 0 are used instead.
This is backwards compatible, as earlier versions of json-c are
using the same integer values in their present definitions.

Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
---
 libdmmp/libdmmp_private.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/libdmmp/libdmmp_private.h b/libdmmp/libdmmp_private.h
index ac85b63f..4378962b 100644
--- a/libdmmp/libdmmp_private.h
+++ b/libdmmp/libdmmp_private.h
@@ -82,7 +82,7 @@ static out_type func_name(struct dmmp_context *ctx, const char *var_name) { \
 do { \
 	json_type j_type = json_type_null; \
 	json_object *j_obj_tmp = NULL; \
-	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != TRUE) { \
+	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != 1) { \
 		_error(ctx, "Invalid JSON output from multipathd IPC: " \
 		       "key '%s' not found", key); \
 		rc = DMMP_ERR_IPC_ERROR; \
-- 
2.17.2

--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel

Re: [PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Martin Wilck]

Hi Ben,

On Thu, 2020-05-28 at 23:10 -0500, Benjamin Marzinski wrote:
> From: Björn Esser <besser82@fedoraproject.org>
> 
> TRUE/FALSE are not defined anymore.  1 and 0 are used instead.
> This is backwards compatible, as earlier versions of json-c are
> using the same integer values in their present definitions.
> 
> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
> ---
>  libdmmp/libdmmp_private.h | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libdmmp/libdmmp_private.h b/libdmmp/libdmmp_private.h
> index ac85b63f..4378962b 100644
> --- a/libdmmp/libdmmp_private.h
> +++ b/libdmmp/libdmmp_private.h
> @@ -82,7 +82,7 @@ static out_type func_name(struct dmmp_context *ctx,
> const char *var_name) { \
>  do { \
>  	json_type j_type = json_type_null; \
>  	json_object *j_obj_tmp = NULL; \
> -	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != TRUE)
> { \
> +	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != 1) { \
>  		_error(ctx, "Invalid JSON output from multipathd IPC: "
> \
>  		       "key '%s' not found", key); \
>  		rc = DMMP_ERR_IPC_ERROR; \

Did you see 
https://www.redhat.com/archives/dm-devel/2020-May/msg00261.html ?

This has first been reported to the list by Christian (
https://www.redhat.com/archives/dm-devel/2020-April/msg00261.html), and
brought to my attention later by Xose. I personally thought the change
from boolean to int is a step in the wrong direction, therefore I
submitted my modified version using stdboolh. If everyone else is fine
with the int, it's not worth arguing about it.

Regards,
Martin

PS: Can anyone explain why json-c did this? Looks like a "cause hassle
for downstream devs and users for no good reason" kind of thing to
me...

-- 
Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107
SUSE  Software Solutions Germany GmbH
HRB 36809, AG Nürnberg GF: Felix
Imendörffer

Re: [PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Xose Vazquez Perez]

On 5/29/20 11:12 AM, Martin Wilck wrote:

> On Thu, 2020-05-28 at 23:10 -0500, Benjamin Marzinski wrote:
>> From: Björn Esser <besser82@fedoraproject.org>
>>
>> TRUE/FALSE are not defined anymore.  1 and 0 are used instead.
>> This is backwards compatible, as earlier versions of json-c are
>> using the same integer values in their present definitions.
>>
>> Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
>> ---
>>   libdmmp/libdmmp_private.h | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/libdmmp/libdmmp_private.h b/libdmmp/libdmmp_private.h
>> index ac85b63f..4378962b 100644
>> --- a/libdmmp/libdmmp_private.h
>> +++ b/libdmmp/libdmmp_private.h
>> @@ -82,7 +82,7 @@ static out_type func_name(struct dmmp_context *ctx,
>> const char *var_name) { \
>>   do { \
>>   	json_type j_type = json_type_null; \
>>   	json_object *j_obj_tmp = NULL; \
>> -	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != TRUE)
>> { \
>> +	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != 1) { \
>>   		_error(ctx, "Invalid JSON output from multipathd IPC: "
>> \
>>   		       "key '%s' not found", key); \
>>   		rc = DMMP_ERR_IPC_ERROR; \
> 
> Did you see
> https://www.redhat.com/archives/dm-devel/2020-May/msg00261.html ?
> 
> This has first been reported to the list by Christian (
> https://www.redhat.com/archives/dm-devel/2020-April/msg00261.html), and
> brought to my attention later by Xose. I personally thought the change
> from boolean to int is a step in the wrong direction, therefore I
> submitted my modified version using stdboolh. If everyone else is fine
> with the int, it's not worth arguing about it.
> 
> Regards,
> Martin
> 
> PS: Can anyone explain why json-c did this? Looks like a "cause hassle
> for downstream devs and users for no good reason" kind of thing to
> me...
> 

Add Eric Haszlakiewicz to CC.

Done at: https://github.com/json-c/json-c/commit/0992aac61f8b087efd7094e9ac2b84fa9c040fcd

Re: [PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Benjamin Marzinski]

On Fri, May 29, 2020 at 09:12:30AM +0000, Martin Wilck wrote:
> Hi Ben,
> 
> On Thu, 2020-05-28 at 23:10 -0500, Benjamin Marzinski wrote:
> > From: Björn Esser <besser82@fedoraproject.org>
> > 
> > TRUE/FALSE are not defined anymore.  1 and 0 are used instead.
> > This is backwards compatible, as earlier versions of json-c are
> > using the same integer values in their present definitions.
> > 
> > Signed-off-by: Benjamin Marzinski <bmarzins@redhat.com>
> > ---
> >  libdmmp/libdmmp_private.h | 2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> > 
> > diff --git a/libdmmp/libdmmp_private.h b/libdmmp/libdmmp_private.h
> > index ac85b63f..4378962b 100644
> > --- a/libdmmp/libdmmp_private.h
> > +++ b/libdmmp/libdmmp_private.h
> > @@ -82,7 +82,7 @@ static out_type func_name(struct dmmp_context *ctx,
> > const char *var_name) { \
> >  do { \
> >  	json_type j_type = json_type_null; \
> >  	json_object *j_obj_tmp = NULL; \
> > -	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != TRUE)
> > { \
> > +	if (json_object_object_get_ex(j_obj, key, &j_obj_tmp) != 1) { \
> >  		_error(ctx, "Invalid JSON output from multipathd IPC: "
> > \
> >  		       "key '%s' not found", key); \
> >  		rc = DMMP_ERR_IPC_ERROR; \
> 
> Did you see 
> https://www.redhat.com/archives/dm-devel/2020-May/msg00261.html ?

Nope. Overlooked it. My bad. I'm fine with any version that compiles.
Either stdbool or int is fine.

-Ben

> 
> This has first been reported to the list by Christian (
> https://www.redhat.com/archives/dm-devel/2020-April/msg00261.html), and
> brought to my attention later by Xose. I personally thought the change
> from boolean to int is a step in the wrong direction, therefore I
> submitted my modified version using stdboolh. If everyone else is fine
> with the int, it's not worth arguing about it.
> 
> Regards,
> Martin
> 
> PS: Can anyone explain why json-c did this? Looks like a "cause hassle
> for downstream devs and users for no good reason" kind of thing to
> me...
> 
> -- 
> Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107
> SUSE  Software Solutions Germany GmbH
> HRB 36809, AG Nürnberg GF: Felix
> Imendörffer
> 

Re: [PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Martin Wilck]

On Fri, 2020-05-29 at 11:19 -0500, Benjamin Marzinski wrote:
> On Fri, May 29, 2020 at 09:12:30AM +0000, Martin Wilck wrote:
> > 
> > Did you see 
> > https://www.redhat.com/archives/dm-devel/2020-May/msg00261.html ?
> 
> Nope. Overlooked it. My bad. I'm fine with any version that compiles.
> Either stdbool or int is fine.

Ok. I'll take the liberty to apply mine on my upstream-queue branch,
then. Let Christophe have the final say when he merges next time.

Martin

-- 
Dr. Martin Wilck <mwilck@suse.com>, Tel. +49 (0)911 74053 2107
SUSE  Software Solutions Germany GmbH
HRB 36809, AG Nürnberg GF: Felix
Imendörffer

Re: [PATCH 1/1] libdmmp: Add support for upcoming json-c 0.14.0.

[Eric Hawicz]

On 5/29/2020 7:45 AM, Xose Vazquez Perez wrote:
> On 5/29/20 11:12 AM, Martin Wilck wrote:
>
>> Did you see
>> https://www.redhat.com/archives/dm-devel/2020-May/msg00261.html ?
>>
>> This has first been reported to the list by Christian (
>> https://www.redhat.com/archives/dm-devel/2020-April/msg00261.html), and
>> brought to my attention later by Xose. I personally thought the change
>> from boolean to int is a step in the wrong direction, therefore I
>> submitted my modified version using stdboolh. If everyone else is fine
>> with the int, it's not worth arguing about it.
>>
>> Regards,
>> Martin
>>
>> PS: Can anyone explain why json-c did this? Looks like a "cause hassle
>> for downstream devs and users for no good reason" kind of thing to
>> me...
>>
>
> Add Eric Haszlakiewicz to CC.
>
> Done at: 
> https://github.com/json-c/json-c/commit/0992aac61f8b087efd7094e9ac2b84fa9c040fcd


TRUE and FALSE are rather generic defines that json-c had no business 
ever setting, especially to json-c specific values, and the fact that we 
were doing so was conflicting with other cases that already defined it.  
Sorry about the hassle, but I consider that a very good reason to remove 
them.


Eric