[gentoo-dev] Duplicated entries in use.desc and use.local.desc

[gentoo-dev] Duplicated entries in use.desc and use.local.desc

[Mark Loeser]

[-- Attachment #1: Type: text/plain, Size: 873 bytes --]

Well, the subject really says it all.  On bug #89056 it seems to be thought
that the duplication is not necessarily a bad thing.  I'm not sure this is
such a great idea myself, since if the USE flag needs a separate description
for a package, it means that the behaviour from that flag could be different
from package to package.  Currently most of the duplicated flags that I
looked at are not at all different and are just restating the same thing in
two files.

So, what do we want to do about this?  Should we have some repoman warning if
a USE flag exists in both files, or should it be acceptable?

-- 
Mark Loeser   -   Gentoo Developer (cpp gcc-porting qa toolchain x86)
email         -   halcy0n AT gentoo DOT org
                  mark AT halcy0n DOT com
web           -   http://dev.gentoo.org/~halcy0n/
                  http://www.halcy0n.com

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

Re: [gentoo-dev] Duplicated entries in use.desc and use.local.desc

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 533 bytes --]

On Sun, 12 Feb 2006 03:19:25 -0500 Mark Loeser <halcy0n@gentoo.org>
wrote:
| So, what do we want to do about this?  Should we have some repoman
| warning if a USE flag exists in both files, or should it be
| acceptable?

The whole idea of global USE flags is that they're the same for all
packages that use said flag, and thus do not get a use.local.desc
entry.

-- 
Ciaran McCreesh : Gentoo Developer (Wearer of the shiny hat)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

[gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[R Hill]

Ciaran McCreesh wrote:
> On Sun, 12 Feb 2006 03:19:25 -0500 Mark Loeser <halcy0n@gentoo.org>
> wrote:
> | So, what do we want to do about this?  Should we have some repoman
> | warning if a USE flag exists in both files, or should it be
> | acceptable?
> 
> The whole idea of global USE flags is that they're the same for all
> packages that use said flag, and thus do not get a use.local.desc
> entry.

a global USE flag duplicated in use.local.desc could be used to give specific
information about exactly what effect the flag has on a certain package, or if
for some reason it does differ slightly from the global meaning.

global use flags (searching: doc)
************************************************************
[-    ] doc - Adds extra documentation (API, Javadoc, etc)

local use flags (searching: doc)
************************************************************
[-    ] doc (app-examples/fakeapp):
Build user manuals in PDF format (requires ps2pdf)


it's been mentioned before but i don't think anyone jumped on the idea.  i
personally don't care either way, just wanted to bring it up while it's on topic. ;P


--de.

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Simon Stelling]

R Hill wrote:
> a global USE flag duplicated in use.local.desc could be used to give specific
> information about exactly what effect the flag has on a certain package, or if
> for some reason it does differ slightly from the global meaning.
> 
> global use flags (searching: doc)
> ************************************************************
> [-    ] doc - Adds extra documentation (API, Javadoc, etc)
> 
> local use flags (searching: doc)
> ************************************************************
> [-    ] doc (app-examples/fakeapp):
> Build user manuals in PDF format (requires ps2pdf)

That'd be bad practice. When a new global use flag is made, the requirement is
that all local use flags which would get united have *the same meaning*. If the
meaning is the same, it doesn't make sense to mention it twice. If the meaning
differs (slightly or not), it should get a local use flag.

-- 
Simon Stelling
Gentoo/AMD64 Operational Co-Lead
blubb@gentoo.org
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Marius Mauch]

[-- Attachment #1: Type: text/plain, Size: 1675 bytes --]

On Sun, 12 Feb 2006 17:49:26 +0100
Simon Stelling <blubb@gentoo.org> wrote:

> R Hill wrote:
> > a global USE flag duplicated in use.local.desc could be used to
> > give specific information about exactly what effect the flag has on
> > a certain package, or if for some reason it does differ slightly
> > from the global meaning.
> > 
> > global use flags (searching: doc)
> > ************************************************************
> > [-    ] doc - Adds extra documentation (API, Javadoc, etc)
> > 
> > local use flags (searching: doc)
> > ************************************************************
> > [-    ] doc (app-examples/fakeapp):
> > Build user manuals in PDF format (requires ps2pdf)
> 
> That'd be bad practice. When a new global use flag is made, the
> requirement is that all local use flags which would get united have
> *the same meaning*. If the meaning is the same, it doesn't make sense
> to mention it twice. If the meaning differs (slightly or not), it
> should get a local use flag.

IIRC the idea behind duplication was not to use a flag for different
purposes, but have a generic description in use.desc (like "doc: build
additional docs") and give a more detailed description in
use.local.desc (like "doc: build API docs and manual as pdf and html").
See http://thread.gmane.org/gmane.linux.gentoo.portage.devel/618 and
http://thread.gmane.org/gmane.linux.gentoo.devel/26035 for the original
threads about this.

-- 
Public Key at http://www.genone.de/info/gpg-key.pub

In the beginning, there was nothing. And God said, 'Let there be
Light.' And there was still nothing, but you could see a bit better.

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

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Mark Loeser]

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

Marius Mauch <genone@gentoo.org> said:
> IIRC the idea behind duplication was not to use a flag for different
> purposes, but have a generic description in use.desc (like "doc: build
> additional docs") and give a more detailed description in
> use.local.desc (like "doc: build API docs and manual as pdf and html").
> See http://thread.gmane.org/gmane.linux.gentoo.portage.devel/618 and
> http://thread.gmane.org/gmane.linux.gentoo.devel/26035 for the original
> threads about this.

In all of the examples I have seen it seems that the "more detailed
description" doesn't really give that much more useful information.  It just
seems to be duplicating the same information in different words.

Currently the flags I found in both use.desc and use.local.desc are exact
copies, so I will talk to the maintainers for those packages and make sure
they are alright with removing their entry.

On a more global scale, we should decide if this is valid though.  I haven't
exactly been convinced that it is useful, but I'm not opposed to the idea.
I'd just like to see a decision one way or another.

-- 
Mark Loeser   -   Gentoo Developer (cpp gcc-porting qa toolchain x86)
email         -   halcy0n AT gentoo DOT org
                  mark AT halcy0n DOT com
web           -   http://dev.gentoo.org/~halcy0n/
                  http://www.halcy0n.com

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

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Marius Mauch]

[-- Attachment #1: Type: text/plain, Size: 1805 bytes --]

On Sun, 12 Feb 2006 14:49:55 -0500
Mark Loeser <halcy0n@gentoo.org> wrote:

> Marius Mauch <genone@gentoo.org> said:
> > IIRC the idea behind duplication was not to use a flag for different
> > purposes, but have a generic description in use.desc (like "doc:
> > build additional docs") and give a more detailed description in
> > use.local.desc (like "doc: build API docs and manual as pdf and
> > html"). See
> > http://thread.gmane.org/gmane.linux.gentoo.portage.devel/618 and
> > http://thread.gmane.org/gmane.linux.gentoo.devel/26035 for the
> > original threads about this.
> 
> In all of the examples I have seen it seems that the "more detailed
> description" doesn't really give that much more useful information.
> It just seems to be duplicating the same information in different
> words.
>
> Currently the flags I found in both use.desc and use.local.desc are
> exact copies, so I will talk to the maintainers for those packages
> and make sure they are alright with removing their entry.

For the record, I didn't look at the current cases nor do I have a
positions for or against the idea, just providing some general
background info.

> On a more global scale, we should decide if this is valid though.  I
> haven't exactly been convinced that it is useful, but I'm not opposed
> to the idea. I'd just like to see a decision one way or another.

Yeah, unfortunately none of the people proposing the idea ever wrote a
glep or even filed a bug regarding this (AFAIK), also as the quoted
threads show there are a number of possible solutions for this.

Marius

-- 
Public Key at http://www.genone.de/info/gpg-key.pub

In the beginning, there was nothing. And God said, 'Let there be
Light.' And there was still nothing, but you could see a bit better.

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

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Mark Loeser]

[-- Attachment #1: Type: text/plain, Size: 1110 bytes --]

Marius Mauch <genone@gentoo.org> said:
> Yeah, unfortunately none of the people proposing the idea ever wrote a
> glep or even filed a bug regarding this (AFAIK), also as the quoted
> threads show there are a number of possible solutions for this.

It doesn't seem that too many people feel strongly in favor of it (judging by
responses so far), so why don't we just say that it is invalid until such a
time where someone can come up with real-life  examples where this would prove
to be beneficial.  Is this something that repoman could check for (should I
file a bug), or just something to keep my eye out for?

I don't see this as a groundbreaking change that requires a GLEP or anything,
especially since I eliminated most of the duplicates today, after talking
with the respective maintainers.  There are probably only 3 or 4 left.

-- 
Mark Loeser   -   Gentoo Developer (cpp gcc-porting qa toolchain x86)
email         -   halcy0n AT gentoo DOT org
                  mark AT halcy0n DOT com
web           -   http://dev.gentoo.org/~halcy0n/
                  http://www.halcy0n.com

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

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Ciaran McCreesh]

[-- Attachment #1: Type: text/plain, Size: 580 bytes --]

On Sun, 12 Feb 2006 22:19:29 -0500 Mark Loeser <halcy0n@gentoo.org>
wrote:
| I don't see this as a groundbreaking change that requires a GLEP or
| anything, especially since I eliminated most of the duplicates today,
| after talking with the respective maintainers.  There are probably
| only 3 or 4 left.

The original "someone should GLEP this" was on adding extended USE flag
descriptions to metadata.xml.

-- 
Ciaran McCreesh : Gentoo Developer (Wearer of the shiny hat)
Mail            : ciaranm at gentoo.org
Web             : http://dev.gentoo.org/~ciaranm


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

[gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[R Hill]

Marius Mauch wrote:
> On Sun, 12 Feb 2006 14:49:55 -0500
> Mark Loeser <halcy0n@gentoo.org> wrote:
>> On a more global scale, we should decide if this is valid though.  I
>> haven't exactly been convinced that it is useful, but I'm not opposed
>> to the idea. I'd just like to see a decision one way or another.
> 
> Yeah, unfortunately none of the people proposing the idea ever wrote a
> glep or even filed a bug regarding this (AFAIK), also as the quoted
> threads show there are a number of possible solutions for this.

TGL did some work on this under bug #84884, though his changes are more invasive 
than what i had in mind.  I don't see the need for portage to dig through 
use.*desc when euse already works and equery can pretty easily be made to.

Anyways I just like anything that makes use.desc more useful than

foo - adds support for foo


--de.

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Mark Loeser]

[-- Attachment #1: Type: text/plain, Size: 557 bytes --]

R Hill <dirtyepic.sk@gmail.com> said:
> Anyways I just like anything that makes use.desc more useful than
> 
> foo - adds support for foo

That's really a completely separate issue.  By allowing duplicate entries we
just allow people to put useless information in two places instead of one.

-- 
Mark Loeser   -   Gentoo Developer (cpp gcc-porting qa toolchain x86)
email         -   halcy0n AT gentoo DOT org
                  mark AT halcy0n DOT com
web           -   http://dev.gentoo.org/~halcy0n/
                  http://www.halcy0n.com

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

[gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[R Hill]

Mark Loeser wrote:
> R Hill <dirtyepic.sk@gmail.com> said:
>> Anyways I just like anything that makes use.desc more useful than
>>
>> foo - adds support for foo
> 
> That's really a completely separate issue.  By allowing duplicate entries we
> just allow people to put useless information in two places instead of one.

Alright then.  Seems reasonable by me.

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Thomas de Grenier de Latour]

Le Sun, 12 Feb 2006 23:28:05 -0500,
Mark Loeser <halcy0n@gentoo.org> a écrit :

> By allowing duplicate entries we just allow people to put useless
> information in two places instead of one.
> 

Maybe i'm a bit naive, but that sounds very pessimistic to me. I would
rather think that devs who will add a use.local.desc entry for a global
flag and their package will only do it when they have something more
useful than the default "Adds foobar support" to say. Otherwise, why
would they do so?

--
TGL.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Re: Duplicated entries in use.desc and use.local.desc

[Thomas de Grenier de Latour]

Le Sun, 12 Feb 2006 21:39:22 -0600,
R Hill <dirtyepic.sk@gmail.com> a écrit :

> TGL did some work on this under bug #84884, though his changes are
> more invasive than what i had in mind. I don't see the need for
> portage to dig through use.*desc when euse already works and equery
> can pretty easily be made to.

If this "special" USE descriptions (the one in use.local.desc when the
flag is also global) are allowed, then it's in "emerge -pv" output
that displaying them is the most useful. Nobody wants to manually call
euses for each package he's about to emerge/update just in case one of
the well known flags they use has a special description. That's
something that should simply come to his attention when it's the case,
it's much easier this way.

IIRC, the behavior of my patch was that when the "--use-desc-special"
option was used, and some packages/flags in the list had special
descriptions, the relevant informations were displayed at the end of 
the usual output:

  % emerge -puvD --use-desc-special world
  ...
  [ebuild  U  ] net-ftp/pure-ftpd-1.0.20-r2 -caps -ldap mysql pam
  -postgres ssl -vchroot
  [ebuild  U  ] ...
  ...
  These USE flags have a package-specific description:
    pure-ftpd:mysql - Allow storing accounts infos in a MySQL DB
  ...

Note that this patch doesn't makes portage diging through use.*desc when
this option is not used.

As for the two other patches (repoman and equery), it was just some code
cleanup (remove their own duplicate implementation of use*.desc parsers,
to replace it with some shared code).

> Anyways I just like anything that makes use.desc more useful than
> 
> foo - adds support for foo

In many cases, you just can't give a better description for a global
flag, because it has two much different purposes depending on the 
context (the package using it).

Take the "mysql" flag, i think it's a typical example of global flag 
with different meanings: many users will enable it thinking of the PHP
bindings, whereas they don't care about using a MySQL DB to store 
their FTP accounts or their music collection metadatas.

Or even take some less widely used flags, like "sqlite3"; on just six
packages using it, it can be:
 - add sqlite support (which happens to be v3 only)
 - add support for sqlite3 (may be in addition to the v2 controlled by 
   the "sqlite" flag)
 - use sqlite3 for backend (but v2 has priority if "sqlite" is enabled)
 - use sqlite3 for backend (and die if "sqlite" is enabled too)
Again, the global description ("Adds support for sqlite3") is vague 
enough to seem ~correct in all cases, but actually gives no clue about
what turning on the flag means.

--
TGL.
-- 
gentoo-dev@gentoo.org mailing list