empty strings

empty strings

[Frans Meulenbroeks]

triggered by a remark on irc, I decided to do a grep on "" in all our recipes.
That gave some interesting observations. I've fixed a few things from
the type X += "" or X_append = ""
See the patch set I just send.

However there are also a few things that are not so obvious and for
which I want to ask your opinion, Some seem useless, others are
probably needed, but wanted to raise all issues that could be
unneeded.
Note that the recipes are just examples, ofthen there are more recipes
that have the same pattern.

libvorbis/libvorbis_1.2.3.bb:EXTRA_OECONF = ""
aircrack/aircrack_2.1.bb:DEPENDS = ""
apt/apt-native.inc:PACKAGES = ""  (this is a native recipe, isn't this
always the case for native recipes ? )
bind/bind_9.3.1.bb:RPEDENDS = "" (this one has a DEPENDS on openssl, I
would expect it also to rdepend on openssl in this case)
c3110x/cx3110x_1.1.bb:export LDFLAGS = ""
classpath/classpath.inc:RPROVIDES_${PN} = ""
fuse/fuse-module_2.7.4.bb:      LDFLAGS=""
glibc/glibc_2.10.1.bb:  CPPFLAGS="" oe_runconf (I was wondering of the
resetting CPPFLAGS is useful, didn't really dig into this one)
gs/gs-fonts-std_8.11.bb:HOMEPAGE = ""
gsoap/gsoap-native_2.7.13.bb:EXTRA_OEMAKE = ""
ipkg/ipkg-native.inc:PROVIDES = ""  (hm, no provides)
meta/staging-linkage_1.0.bb:SRC_URI = ""
psmisc/psmisc.inc:FILES_${PN} = "" (guess all files are added to other
packages, if that is so, should there be a line like this?)
tmake/tmake_1.11.bb:FILES = ""
ttf-fonts/ttf-hunkyfonts_0.3.0.bb:RRECOMMENDS_${PN}-dbg = "" (seems
somewhat strange to me)


this one also looked weird; guess it is ok but found it odd that these are reset
hplip/hplip_2.8.10.bb:EXTRA_OEMAKE = 'BUILD_SYS="" HOST_SYS=""'

I would like opinions on which of these constructs is useful and which
not. (and yes, I know an assingment may be used to nullify an earlier
assingment in an .inc file and I am also aware of cases where there is
something like X_virtclass_native = "" which is used to nullify
non-native assignments to X.

Enjoy, Frans

Re: empty strings

[Frans Meulenbroeks]

ping.

people don't care?
I feel there are some issues in the examples below (and some easy
cleanup actions).

Frans

2010/10/14 Frans Meulenbroeks <fransmeulenbroeks@gmail.com>:
> triggered by a remark on irc, I decided to do a grep on "" in all our recipes.
> That gave some interesting observations. I've fixed a few things from
> the type X += "" or X_append = ""
> See the patch set I just send.
>
> However there are also a few things that are not so obvious and for
> which I want to ask your opinion, Some seem useless, others are
> probably needed, but wanted to raise all issues that could be
> unneeded.
> Note that the recipes are just examples, ofthen there are more recipes
> that have the same pattern.
>
> libvorbis/libvorbis_1.2.3.bb:EXTRA_OECONF = ""
> aircrack/aircrack_2.1.bb:DEPENDS = ""
> apt/apt-native.inc:PACKAGES = ""  (this is a native recipe, isn't this
> always the case for native recipes ? )
> bind/bind_9.3.1.bb:RPEDENDS = "" (this one has a DEPENDS on openssl, I
> would expect it also to rdepend on openssl in this case)
> c3110x/cx3110x_1.1.bb:export LDFLAGS = ""
> classpath/classpath.inc:RPROVIDES_${PN} = ""
> fuse/fuse-module_2.7.4.bb:      LDFLAGS=""
> glibc/glibc_2.10.1.bb:  CPPFLAGS="" oe_runconf (I was wondering of the
> resetting CPPFLAGS is useful, didn't really dig into this one)
> gs/gs-fonts-std_8.11.bb:HOMEPAGE = ""
> gsoap/gsoap-native_2.7.13.bb:EXTRA_OEMAKE = ""
> ipkg/ipkg-native.inc:PROVIDES = ""  (hm, no provides)
> meta/staging-linkage_1.0.bb:SRC_URI = ""
> psmisc/psmisc.inc:FILES_${PN} = "" (guess all files are added to other
> packages, if that is so, should there be a line like this?)
> tmake/tmake_1.11.bb:FILES = ""
> ttf-fonts/ttf-hunkyfonts_0.3.0.bb:RRECOMMENDS_${PN}-dbg = "" (seems
> somewhat strange to me)
>
>
> this one also looked weird; guess it is ok but found it odd that these are reset
> hplip/hplip_2.8.10.bb:EXTRA_OEMAKE = 'BUILD_SYS="" HOST_SYS=""'
>
> I would like opinions on which of these constructs is useful and which
> not. (and yes, I know an assingment may be used to nullify an earlier
> assingment in an .inc file and I am also aware of cases where there is
> something like X_virtclass_native = "" which is used to nullify
> non-native assignments to X.
>
> Enjoy, Frans
>

Re: empty strings

[Chris Larson]

On Fri, Oct 22, 2010 at 2:00 AM, Frans Meulenbroeks <
fransmeulenbroeks@gmail.com> wrote:

> ping.
>
> people don't care?
> I feel there are some issues in the examples below (and some easy
> cleanup actions).
>

Feel free to use my ack for the series, the ones you've sent patches for
seem pretty obviously correct to me.  The other empty strings may need more
review.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics

Re: empty strings

[Frans Meulenbroeks]

2010/10/22 Chris Larson <clarson@kergoth.com>:
> On Fri, Oct 22, 2010 at 2:00 AM, Frans Meulenbroeks <
> fransmeulenbroeks@gmail.com> wrote:
>
>> ping.
>>
>> people don't care?
>> I feel there are some issues in the examples below (and some easy
>> cleanup actions).
>>
>
> Feel free to use my ack for the series, the ones you've sent patches for
> seem pretty obviously correct to me.  The other empty strings may need more
> review.

pushed the patches posted before.
also fixed (as a starter) the EXTRA_OECONF = "" things
(ran bitbake -e before and after and compared the differences)
Might peek into some other things later

There are still some EXTRA_OECONF_virtclass-native = "" in glib-2.0;
we can eliminate this once we have the -target suffix
(and rename EXTRA_OECONF to EXTRA_OECONF-target).
tthen again: glib-2.0 with 10 recipes (and a native one) could perhaps
use some pruning)

Frans

Re: empty strings

[Robert Schuster]

[-- Attachment #1: Type: text/plain, Size: 400 bytes --]

Am 22.10.2010 11:00, schrieb Frans Meulenbroeks:
>> classpath/classpath.inc:RPROVIDES_${PN} = ""
It would not be a good idea to blindly remove this. The effect of this
can only be seen at runtime. I don't have the time to look into this
deeply now but we definitely did this because we don't like the default
RPROVIDES that OE sets for classpath and classpath-minimal.

Regards
Robert




[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 270 bytes --]

Re: empty strings

[Khem Raj]

On Fri, Oct 22, 2010 at 2:00 AM, Frans Meulenbroeks
<fransmeulenbroeks@gmail.com> wrote:
> ping.
>
> people don't care?
> I feel there are some issues in the examples below (and some easy
> cleanup actions).

If you dont generalize and pick and test a recipe at a time that would
be much sane way to go with such changes.
Sometimes they might look obvious but they may not be. So my humble
request is to go recipe by recipe instead of
some seding and grepping.

>
> Frans
>
> 2010/10/14 Frans Meulenbroeks <fransmeulenbroeks@gmail.com>:
>> triggered by a remark on irc, I decided to do a grep on "" in all our recipes.
>> That gave some interesting observations. I've fixed a few things from
>> the type X += "" or X_append = ""
>> See the patch set I just send.
>>
>> However there are also a few things that are not so obvious and for
>> which I want to ask your opinion, Some seem useless, others are
>> probably needed, but wanted to raise all issues that could be
>> unneeded.
>> Note that the recipes are just examples, ofthen there are more recipes
>> that have the same pattern.
>>
>> libvorbis/libvorbis_1.2.3.bb:EXTRA_OECONF = ""
>> aircrack/aircrack_2.1.bb:DEPENDS = ""
>> apt/apt-native.inc:PACKAGES = ""  (this is a native recipe, isn't this
>> always the case for native recipes ? )
>> bind/bind_9.3.1.bb:RPEDENDS = "" (this one has a DEPENDS on openssl, I
>> would expect it also to rdepend on openssl in this case)
>> c3110x/cx3110x_1.1.bb:export LDFLAGS = ""
>> classpath/classpath.inc:RPROVIDES_${PN} = ""
>> fuse/fuse-module_2.7.4.bb:      LDFLAGS=""
>> glibc/glibc_2.10.1.bb:  CPPFLAGS="" oe_runconf (I was wondering of the
>> resetting CPPFLAGS is useful, didn't really dig into this one)
>> gs/gs-fonts-std_8.11.bb:HOMEPAGE = ""
>> gsoap/gsoap-native_2.7.13.bb:EXTRA_OEMAKE = ""
>> ipkg/ipkg-native.inc:PROVIDES = ""  (hm, no provides)
>> meta/staging-linkage_1.0.bb:SRC_URI = ""
>> psmisc/psmisc.inc:FILES_${PN} = "" (guess all files are added to other
>> packages, if that is so, should there be a line like this?)
>> tmake/tmake_1.11.bb:FILES = ""
>> ttf-fonts/ttf-hunkyfonts_0.3.0.bb:RRECOMMENDS_${PN}-dbg = "" (seems
>> somewhat strange to me)
>>
>>
>> this one also looked weird; guess it is ok but found it odd that these are reset
>> hplip/hplip_2.8.10.bb:EXTRA_OEMAKE = 'BUILD_SYS="" HOST_SYS=""'
>>
>> I would like opinions on which of these constructs is useful and which
>> not. (and yes, I know an assingment may be used to nullify an earlier
>> assingment in an .inc file and I am also aware of cases where there is
>> something like X_virtclass_native = "" which is used to nullify
>> non-native assignments to X.
>>
>> Enjoy, Frans
>>
>
> _______________________________________________
> Openembedded-devel mailing list
> Openembedded-devel@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-devel
>

Re: empty strings

[Frans Meulenbroeks]

2010/10/25 Khem Raj <raj.khem@gmail.com>:
> On Fri, Oct 22, 2010 at 2:00 AM, Frans Meulenbroeks
> <fransmeulenbroeks@gmail.com> wrote:
>> ping.
>>
>> people don't care?
>> I feel there are some issues in the examples below (and some easy
>> cleanup actions).
>
> If you dont generalize and pick and test a recipe at a time that would
> be much sane way to go with such changes.
> Sometimes they might look obvious but they may not be. So my humble
> request is to go recipe by recipe instead of
> some seding and grepping.
>

I will.

Actually one of the reasons for posting this was that someone perhaps
could say upfront something like:
"Nah this is sound, no need to waste time on this" (as actually done
by Robert in a previous email"

The other reason was that I hoped that some people would be triggered
to peek in their recipes and resolev things themselves.

And the 3rd reason was to get some feedback on things that look
suspicious or strange.
E.g. these ones:

apt/apt-native.inc:PACKAGES = ""  (this is a native recipe, isn't this
always the case for native recipes ? )
bind/bind_9.3.1.bb:RPEDENDS = "" (this one has a DEPENDS on openssl, I
would expect it also to rdepend on openssl in this case)
glibc/glibc_2.10.1.bb:  CPPFLAGS="" oe_runconf (I was wondering of the
resetting CPPFLAGS is useful, didn't really dig into this one)
ipkg/ipkg-native.inc:PROVIDES = ""  (hm, no provides)
psmisc/psmisc.inc:FILES_${PN} = "" (guess all files are added to other
packages, if that is so, should there be a line like this?)
tmake/tmake_1.11.bb:FILES = ""
ttf-fonts/ttf-hunkyfonts_0.3.0.bb:RRECOMMENDS_${PN}-dbg = "" (seems
somewhat strange to me)

Frans.