[gentoo-dev] Difference of global/local useflags

[gentoo-dev] Difference of global/local useflags

[Markus Nigbur]

[-- Attachment #1: Type: text/plain, Size: 742 bytes --]

As this question just arised within the german conspiracy and we
couldn't tell exactly, i take this question up to the list:

Why do we distinguish between global and local useflags at all? Is there
a real difference between them?

It's obvious when a useflag should be local (jsut for one package) or
global (some pkgs use it), but what do we gain by this split?

Before someone replies with "you can comment on the actual influence of
a local useflag on the package on the local useflag's deascription" - if
that's the reason, why shouldn't this be useful for the influence of
global useflags on a package?

Regards,
    Markus Nigbur


-- 
 (o_  Markus Nigbur
 //\  Gentoo Linux Developer
[ ]/_ http://www.gentoo.org

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Drake Wyrm]

[-- Attachment #1: Type: text/plain, Size: 977 bytes --]

At 2004-05-20T21:12:06+0200, Markus Nigbur <pyrania@gentoo.org> wrote:
> Why do we distinguish between global and local useflags at all? Is
> there a real difference between them?
> 
> It's obvious when a useflag should be local (jsut for one package) or
> global (some pkgs use it), but what do we gain by this split?
> 
> Before someone replies with "you can comment on the actual influence
> of a local useflag on the package on the local useflag's deascription"
> - if that's the reason, why shouldn't this be useful for the influence
> of global useflags on a package?

Some time ago, somebody mentioned the possibility of adding local use
descriptions to the metadata.xml format. At the time, the idea was a bit
off-topic and was quickly lost in the noise. Has anybody given this idea
any thought?

-- 
Batou: Hey, Major... You ever hear of "human rights"?
Kusanagi: I understand the concept, but I've never seen it in action.
  --Ghost in the Shell

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 661 bytes --]

On Thu, 20 May 2004 12:41:16 -0700 Drake Wyrm <wyrm@haell.com> wrote:
| Some time ago, somebody mentioned the possibility of adding local use
| descriptions to the metadata.xml format. At the time, the idea was a
| bit off-topic and was quickly lost in the noise. Has anybody given
| this idea any thought?

Eh, why not go the whole way and allow any package to provide a more
verbose description of how *any* (local or global) USE flag applies to
the particular package?

-- 
Ciaran McCreesh, Gentoo XMLcracy Member G03X276
(Sparc, MIPS, Vim, si hoc legere scis nimium eruditionis habes)
Mail:    ciaranm at gentoo.org
Web:     http://dev.gentoo.org/~ciaranm


[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 971 bytes --]

On Thu, 2004-05-20 at 21:12, Markus Nigbur wrote:
> Why do we distinguish between global and local useflags at all? Is there
> a real difference between them?

I say, just drop global useflags.
While at it, drop global CFLAGS.
Drop global it's silly.


I think we need to go back to the basics:

What does portage add that native installation doesn't?

* Automatic non-interactive configuration sensitive dependency
resolving.
* Tracking of installed files for easy removing.
* Gentoo integration.

I don't know why useflags was introduced in the first place. But is it
really the best sollution to provide the above?
Can package configuration be stored in the package native format and
still provide dependency data?


I feel that we should try to move away from the monolithic nature of the
portage tree and try to harness the power of "the web" more. To do that
we have to push as much of the package handling as possible upstream.

-John

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

John Nilsson said:
> I feel that we should try to move away from the monolithic nature of the
> portage tree and try to harness the power of "the web" more. To do that
> we have to push as much of the package handling as possible upstream.

Many handle it upstream, i can think of perl and oo.o off the top of my
head, but do you want to go through their customizing ways (aka, the
./configure script) and lose portage's ablity to be run non-interactivly?



-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[Stuart Herbert]

[-- Attachment #1: Type: text/plain, Size: 639 bytes --]

On Thursday 20 May 2004 20:56, Ciaran McCreesh wrote:
> Eh, why not go the whole way and allow any package to provide a more
> verbose description of how *any* (local or global) USE flag applies to
> the particular package?

+1

Makes sense to me.

Stu
-- 
Stuart Herbert                                              stuart@gentoo.org
Gentoo Developer                                       http://www.gentoo.org/
                                                   http://stu.gnqs.org/diary/

GnuPG key id# F9AFC57C available from http://pgp.mit.edu
Key fingerprint = 31FB 50D4 1F88 E227 F319  C549 0C2F 80BA F9AF C57C
--

[-- Attachment #2: signature --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 774 bytes --]

On Thu, 2004-05-20 at 23:06, Joseph Booker wrote:
> John Nilsson said:
> > I feel that we should try to move away from the monolithic nature of the
> > portage tree and try to harness the power of "the web" more. To do that
> > we have to push as much of the package handling as possible upstream.
> 
> Many handle it upstream, i can think of perl and oo.o off the top of my
> head, but do you want to go through their customizing ways (aka, the
> ./configure script) and lose portage's ablity to be run non-interactivly?
> 
> 

I would be fine with something like the kernels "make oldconfig" each
emerge instance.
That would be just as "non-interactive" as:
emerge -uUD world -pv && vim /etc/make.conf && vim /etc/portage/* &&
emerge -uUD world.

-john

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

John Nilsson said:
> On Thu, 2004-05-20 at 23:06, Joseph Booker wrote:
>> John Nilsson said:
>> > I feel that we should try to move away from the monolithic nature of
>> the
>> > portage tree and try to harness the power of "the web" more. To do
>> that
>> > we have to push as much of the package handling as possible upstream.
>>
>> Many handle it upstream, i can think of perl and oo.o off the top of my
>> head, but do you want to go through their customizing ways (aka, the
>> ./configure script) and lose portage's ablity to be run
>> non-interactivly?
>>
>>
>
> I would be fine with something like the kernels "make oldconfig" each
> emerge instance.
> That would be just as "non-interactive" as:
> emerge -uUD world -pv && vim /etc/make.conf && vim /etc/portage/* &&
> emerge -uUD world.

I am sorry, but I don't understand. Your original post recommended the
gentoo devs just dump all their work and go back to portage when it was
much more simplified. Getting rid of all global settings sorta gets rid of
the use of a make.conf, and if there are no use flags, all the settings is
pretty much what do you want masked/unmasked in /etc/portage........an
interesting way of taking a step backwards. Anyways, unless there is some
central configuration for all of portage that handles everything then
mentioing `make oldconfig` doens't make sense, neither does your commands
of updating, editing the configs (for what?) and re-updating (which does
nothing unless you've done an emerge sync)

that and your using the -U flag, not recommend btw

-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 699 bytes --]

On Thu, 2004-05-20 at 22:52, John Nilsson wrote:
> I feel that we should try to move away from the monolithic nature of the
> portage tree and try to harness the power of "the web" more. To do that
> we have to push as much of the package handling as possible upstream.
> 
> -John

Thinking of this some more...

All this metadata, and dependency stuff is really suitable for some web
standardization. It isn't really my area, but wasn't RDF created for
this kind of problems?

By the way, I really urge people to read Roy Fielding's dissertation
about the web architecture. Not entirely related but a good read.
<http://www.ics.uci.edu/~fielding/pubs/dissertation/top.htm>

-John

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

John Nilsson said:
> All this metadata, and dependency stuff is really suitable for some web
> standardization. It isn't really my area, but wasn't RDF created for
> this kind of problems?

I believe that there have been discussions about using xml-based files for
a portage backend, but the current system of using ebuilds works, which is
what matters (and I'm sure theres other arguments depending how you want
to implement it)

>From http://www.w3.org/RDF/:
 The Resource Description Framework (RDF) integrates a variety of
applications from library catalogs and world-wide directories to
syndication and aggregation of news, software, and content to personal
collections of music, photos, and events using XML as an interchange
syntax. The RDF specifications provide a lightweight ontology system to
support the exchange of knowledge on the Web.

So, perfect for RSS feeds and your media libary or playlist or whatever
your media player callls it, but not for something big and complex such as
all the ebuilds (all the ebuilds in one file seems scary and impossible
anyways)

also, maybe you can elaborate on what you mean by web standardization?

-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[Lars Weiler]

[-- Attachment #1: Type: text/plain, Size: 563 bytes --]

* Markus Nigbur <pyrania@gentoo.org> [04/05/20 21:12 +0200]:
> It's obvious when a useflag should be local (jsut for one package) or
> global (some pkgs use it), but what do we gain by this split?

Not to mention the bunch use flags which where local
originally but could be used now by more than one ebuild.

There are furthermore some use-flags which seem to be
similar to me and should be combined to one.

We have currently about 600 use-flags!  Not to mention those
which are not listed in use.desc or use.local.desc.  Time
for tidy up a bit.

Regards, Lars

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 1184 bytes --]

On Thu, 2004-05-20 at 23:38, Joseph Booker wrote:
> I am sorry, but I don't understand. Your original post recommended the
> gentoo devs just dump all their work and go back to portage when it was
> much more simplified.
No not dump, refractor it into compnents, the unix way.

> Getting rid of all global settings sorta gets rid of
> the use of a make.conf, and if there are no use flags, all the settings is
> pretty much what do you want masked/unmasked in /etc/portage........an
> interesting way of taking a step backwards. Anyways, unless there is some
> central configuration for all of portage that handles everything then
> mentioing `make oldconfig` doens't make sense,

The idea being that I configure the package in its native way. Then
portage takes care of the configuration and stores it for updating and
such. Much like what /etc/portage/* does.

>  neither does your commands
> of updating, editing the configs (for what?) and re-updating (which does
> nothing unless you've done an emerge sync)

the first command being -pv to check what needs to be done.
> that and your using the -U flag, not recommend btw
I like to bleed... =)

-John


[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 874 bytes --]

On Thu, 2004-05-20 at 23:44, Joseph Booker wrote:
> John Nilsson said:
> > All this metadata, and dependency stuff is really suitable for some web
> > standardization. It isn't really my area, but wasn't RDF created for
> > this kind of problems?

> So, perfect for RSS feeds and your media libary or playlist or whatever
> your media player callls it, but not for something big and complex such as
> all the ebuilds (all the ebuilds in one file seems scary and impossible
> anyways)
> 
> also, maybe you can elaborate on what you mean by web standardization?

I meant for upstream developers. So instead of ./README or ./INSTALL
stating dependencies ./METADATA.xml or what have you.

A package should provide a uri interface to this information so that
webservices for dependency resolving can be created.

Make portage more of a webservice...

-John

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

John Nilsson said:
> No not dump, refractor it into compnents, the unix way.

as i understand it, the unix way is to make bunchs of small programs
piping to each other. not exactly an elagent thing for portage to do,
`curl http://blah/my-package` | emerge - > /exports/sys_images` seems like
something that would work in such a model, but i dont think portage was
ever meant for such situations

> The idea being that I configure the package in its native way. Then
> portage takes care of the configuration and stores it for updating and
> such. Much like what /etc/portage/* does.

this increases what the user must do to maintain the system, by quite
alot, having portage handle it is really a great conveince

> the first command being -pv to check what needs to be done.
ok, i did not see that

>> that and your using the -U flag, not recommend btw
> I like to bleed... =)

I am so bleeding-edge i get covered in blood everytime i boot gentoo
(love-sources, ~arch in make.conf, udev, breakmygentoo, you name it, ive
either got it or tried it), but -U is replased by
/etc/portage/packages.keywords, with there being quite some advantages to
using the /etc/portage files rather then emerge -U

if it helps to know, /etc/portage/* is more bleeding edge ;)

I sorta get what you mean now, i hope
-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

John Nilsson said:
> I meant for upstream developers. So instead of ./README or ./INSTALL
> stating dependencies ./METADATA.xml or what have you.
>
> A package should provide a uri interface to this information so that
> webservices for dependency resolving can be created.
>
> Make portage more of a webservice...
>
> -John
>

there is already such projects i think, althought not from source.

there was a discusion on this list, i think you can search the archives,
about using a xml-based YaST and tring to convince all upstream developers
to use a standard xml configuration file. this sorta seems as hopeless as
that, unless you can generate the ones your talking about from ./configure
scripts. also, i dont get what portage has to do with webservice, what
your describing seems a bit like having metadata about each package and
just downloading the ebuilds and everything from a web server, which
1. puts more load on the servers then having a local tree
2. makes it harder to modify ebuillds
3. makes no sense as it would not improve the current system in anyway
that i can see

no offense, but it seems to me like you've just read a book on web
services  or something, and would like to have everything take advantage
of such technology, but portage was never designed to be like that, and if
you think it can be, then just remember: code speaks louder then words :P

-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[Todd Berman]

[-- Attachment #1: Type: text/plain, Size: 1026 bytes --]

On Thu, 2004-20-05 at 18:28 -0500, Joseph Booker wrote:
<lots of snipage>
> also, i dont get what portage has to do with webservice, what
> your describing seems a bit like having metadata about each package and
> just downloading the ebuilds and everything from a web server, which
> 1. puts more load on the servers then having a local tree
> 2. makes it harder to modify ebuillds
> 3. makes no sense as it would not improve the current system in anyway
> that i can see
> 
> no offense, but it seems to me like you've just read a book on web
> services  or something, and would like to have everything take advantage
> of such technology, but portage was never designed to be like that, and if
> you think it can be, then just remember: code speaks louder then words :P
> 

Honestly, no offense, but it sounds like you did the same thing.

A webservice running on http://127.0.0.1:123456/ is still a webservice,
and it has lots of potential advantages that another rpc method might
not have.

--Todd

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Drake Wyrm]

[-- Attachment #1: Type: text/plain, Size: 950 bytes --]

At 2004-05-20T20:56:23+0100, Ciaran McCreesh <ciaranm@gentoo.org> wrote:
> On Thu, 20 May 2004 12:41:16 -0700 Drake Wyrm <wyrm@haell.com> wrote:
> | Some time ago, somebody mentioned the possibility of adding local use
> | descriptions to the metadata.xml format. At the time, the idea was a
> | bit off-topic and was quickly lost in the noise. Has anybody given
> | this idea any thought?
> 
> Eh, why not go the whole way and allow any package to provide a more
> verbose description of how *any* (local or global) USE flag applies to
> the particular package?

Exactly what I was thinking. That way, packages which use global USE
flags in the expected way get the standard message but, for example,
bitchx could point out that the gtk USE flag *disables* the console
client.

-- 
Batou: Hey, Major... You ever hear of "human rights"?
Kusanagi: I understand the concept, but I've never seen it in action.
  --Ghost in the Shell

[-- Attachment #2: Type: application/pgp-signature, Size: 197 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Joseph Booker]

Todd Berman said:
> Honestly, no offense, but it sounds like you did the same thing.
>
> A webservice running on http://127.0.0.1:123456/ is still a webservice,
> and it has lots of potential advantages that another rpc method might
> not have.

For some reason i got this message twice in my gentoo-dev folder and once
in my email, 3 times is a bit weird, anyways.............

I am not advicating or going against web services or rpc methods of any
kind, I was simply saying that portage wasn't designed as such and that i
woudln't be able to image an effective way to make it like that, and if
someone wants to prove me wrong, feel free

-- 
 Joe Booker

--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Difference of global/local useflags

[John Nilsson]

[-- Attachment #1: Type: text/plain, Size: 1969 bytes --]

On Fri, 2004-05-21 at 02:45, Todd Berman wrote:
> On Thu, 2004-20-05 at 18:28 -0500, Joseph Booker wrote:
> <lots of snipage>
> > also, i dont get what portage has to do with webservice, what
> > your describing seems a bit like having metadata about each package and
> > just downloading the ebuilds and everything from a web server, which
> > 1. puts more load on the servers then having a local tree
> > 2. makes it harder to modify ebuillds
> > 3. makes no sense as it would not improve the current system in anyway
> > that i can see
> > 
> > no offense, but it seems to me like you've just read a book on web
> > services  or something, and would like to have everything take advantage
> > of such technology, but portage was never designed to be like that, and if
> > you think it can be, then just remember: code speaks louder then words :P
> > 
> 
> Honestly, no offense, but it sounds like you did the same thing.
> 
> A webservice running on http://127.0.0.1:123456/ is still a webservice,
> and it has lots of potential advantages that another rpc method might
> not have.
> 
> --Todd

To make portage a webservice was an example. The key is "web". Not "web"
as in port:80 service, web as in a web of projects.

The web already exist! The dependency graph of the packages in portage
is a web. To formalize this web into a computer language means links and
uri's. Not making portage listen on port:80.

The gain is scalability. That all people (not only gentoo) could use the
same system for dependency resolving.

Well yea I did read a book. As I said it's a good read. If you didn't
know Roy Fielding was one of the guy designing http. His dissertation
discusses the architectural thoughts(called ReST) behind the design, as
can be seen in rfc2616. It's an eye opener.



I might be on the wrong track here though. The solution could be to make
more use of eclass and autogenerate ebuilds on the fly...

-John

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[foser]

[-- Attachment #1: Type: text/plain, Size: 879 bytes --]

On Thu, 2004-05-20 at 20:56 +0100, Ciaran McCreesh wrote:
> On Thu, 20 May 2004 12:41:16 -0700 Drake Wyrm <wyrm@haell.com> wrote:
> | Some time ago, somebody mentioned the possibility of adding local use
> | descriptions to the metadata.xml format. At the time, the idea was a
> | bit off-topic and was quickly lost in the noise. Has anybody given
> | this idea any thought?
> 
> Eh, why not go the whole way and allow any package to provide a more
> verbose description of how *any* (local or global) USE flag applies to
> the particular package?

Global USE flags should have an universal meaning and do the same thing
everywhere. Giving the same flag (slightly) different meanings in
different packages is only confusing.

It is one of the reasons why there is the global/local seperation :
local USE flags can have one defined meaning per ebuild.

- foser

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Difference of global/local useflags

[Chris Gianelloni]

[-- Attachment #1: Type: text/plain, Size: 629 bytes --]

On Thu, 2004-05-20 at 21:37, Drake Wyrm wrote:
> Exactly what I was thinking. That way, packages which use global USE
> flags in the expected way get the standard message but, for example,
> bitchx could point out that the gtk USE flag *disables* the console
> client.

What's up with that, anyway?  I would think that a USE flag is supposed
to "add" functionality.  Shouldn't the ebuild actually build bitchx
twice?  Once with the text interface and once with gtk?

Then again, I quit using bitchx a while back.

-- 
Chris Gianelloni
Developer
Games/LiveCD Teams
Gentoo Linux

Is your power animal a penguin?

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 189 bytes --]