[gentoo-dev] repoman not complaining when going straight to stable

[gentoo-dev] repoman not complaining when going straight to stable

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 1242 bytes --]

This shouldn't be allowed, should it? The package is keyworded only
~x86, and I'm adding amd64 stable keyword, and repoman doesn't complain.

$ ekeyword x86 netboot-0.10.2.ebuild
netboot-0.10.2.ebuild
--- netboot-0.10.2.ebuild	2011-11-23 02:41:03.000000000 +0100
+++ netboot-0.10.2.ebuild.new	2011-11-26 10:12:19.000000000 +0100
@@ -15 +15 @@
-KEYWORDS="~x86"
+KEYWORDS="x86"
$ ekeyword amd64 netboot-0.10.2.ebuild
netboot-0.10.2.ebuild
--- netboot-0.10.2.ebuild	2011-11-26 10:12:19.000000000 +0100
+++ netboot-0.10.2.ebuild.new	2011-11-26 10:12:25.000000000 +0100
@@ -15 +15 @@
-KEYWORDS="x86"
+KEYWORDS="amd64 x86"
$ repoman manifest && repoman full --without-mask
>>> Creating Manifest for /home/ph/gentoo-x86/sys-boot/netboot

RepoMan scours the neighborhood...
  ebuild.minorsyn               1
   sys-boot/netboot/netboot-0.10.1.ebuild: Unquoted Variable on line: 38
  upstream.workaround           1
   sys-boot/netboot/netboot-0.10.2.ebuild: Upstream parallel compilation
bug (ebuild calls emake -j1 on line: 36)

Note: use --include-dev (-d) to check dependencies for 'dev' profiles

RepoMan sez: "You're only giving me a partial QA payment?
              I'll take it this time, but I'm not happy."


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] repoman not complaining when going straight to stable

[Sergei Trofimovich]

[-- Attachment #1: Type: text/plain, Size: 417 bytes --]

On Sat, 26 Nov 2011 10:15:53 +0100
""Paweł Hajdan, Jr."" <phajdan.jr@gentoo.org> wrote:

> This shouldn't be allowed, should it? The package is keyworded only
> ~x86, and I'm adding amd64 stable keyword, and repoman doesn't complain.

It would need to fetch latest CVS file from server to check against previous state.
I don't think repoman does any network IO safe updating metadata.dtd.

-- 

  Sergei

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-dev] repoman not complaining when going straight to stable

["Paweł Hajdan, Jr."]

[-- Attachment #1: Type: text/plain, Size: 778 bytes --]

On 11/26/11 11:29 AM, Sergei Trofimovich wrote:
> On Sat, 26 Nov 2011 10:15:53 +0100
> ""Paweł Hajdan, Jr."" <phajdan.jr@gentoo.org> wrote:
> 
>> This shouldn't be allowed, should it? The package is keyworded only
>> ~x86, and I'm adding amd64 stable keyword, and repoman doesn't complain.
> 
> It would need to fetch latest CVS file from server to check against previous state.
> I don't think repoman does any network IO safe updating metadata.dtd.

Why does it warn about going straight to stable for _new_ ebuilds? Is it
because it knows locally that it's a new file, and because of that
doesn't need to connect to network? That'd make sense.

Would it be possible to add repoman checks (possibly behind a special
flag) that could use "cvs diff" results?


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 203 bytes --]

Re: [gentoo-dev] repoman not complaining when going straight to stable

[Zac Medico]

On 11/29/2011 04:11 AM, "Paweł Hajdan, Jr." wrote:
> On 11/26/11 11:29 AM, Sergei Trofimovich wrote:
>> On Sat, 26 Nov 2011 10:15:53 +0100
>> ""Paweł Hajdan, Jr."" <phajdan.jr@gentoo.org> wrote:
>>
>>> This shouldn't be allowed, should it? The package is keyworded only
>>> ~x86, and I'm adding amd64 stable keyword, and repoman doesn't complain.
>>
>> It would need to fetch latest CVS file from server to check against previous state.
>> I don't think repoman does any network IO safe updating metadata.dtd.
> 
> Why does it warn about going straight to stable for _new_ ebuilds? Is it
> because it knows locally that it's a new file, and because of that
> doesn't need to connect to network? That'd make sense.

Right, repoman parses the files in the CVS directories to figure that out.

> Would it be possible to add repoman checks (possibly behind a special
> flag) that could use "cvs diff" results?

Sure, we can add an --allow-network option. That would also be useful
for checking SRC_URI in this bug:

  https://bugs.gentoo.org/show_bug.cgi?id=315243

We could also add fine-grained options if you want to enable/disable
specific checks that use the network and not others.
-- 
Thanks,
Zac