On 18/05/14 00:40, Klee Dienes wrote:
> On May 17, 2014, at 3:19 PM, Daniel Lintott <dan...@serverb.co.uk> wrote:
>> I've taken a look at your packaging at agree it looks good. I noted a
>> couple of things that I've picked up since I started packaging…
> 
> Thanks for the feedback!  I took most of your suggestions (comments to follow)
> 
>> debian/rules
>>  - no need to define a .PHONY target
>>  - build flags import and set isn't required as dh9 sets all the
>> required build_flags
> 
> Removed the build flags stuff; thanks.
> 
> Do you have a reference regarding the .PHONY target?  Are you saying that 
> there’s a mechanism by which PHONY targets are no longer necessary?  Or just 
> that since those files aren’t present in the source directory, they’re not 
> needed.  Now that you have me thinking about this, I realize that debhelper 
> probably has a bunch of invisitargets that are never getting captured by 
> .PHONY.  So I took the easy way and agreed with you and removed them all.   
> There is a *lot* of documentation out there that implies that they should be 
> present, though … I wonder if there’s a canonical source of guidance anywhere.

I don't have any reference per se.. other than I've not seen it used
personally... (Doesn't mean I'm right btw) :P

Just having a look the only reference I could find was at the beginning
of the following message:
https://www.mail-archive.com/pkg-clamav-devel@lists.alioth.debian.org/msg02836.html

>> debian/control
>>  - section can be removed from libiniparser0 binary package
> 
> I know it’s redundant, but it pains me to have all the other binary packages 
> specify a Section: and not libiniparser0.

Ack... I noticed that it was flagged up by Lintian, so thought I'd
suggest it. I'm not sure what the official view is.

>> debian/watch (Attached)
>>  - Look at GitHub tags directly (not via deprecated github redir)
>>  - Add dversionmangle to remove date/git suffix
> 
> Thanks!  I wonder if it would be possible to add a note to 
> githubredir.debian.net explaining that it’s deprecated and giving the new 
> syntax.  

That may well be possible... Just looking at changelog for the
devscripts package which contains uscan it contains the following:

    + Update the GitHub example.  Based on a patch by YABUKI Yukiharu.
      (Closes: #728182)

in version 2.14.0 (which is present in testing)

> I also wonder if there’s any lintian checks that already look for deprecated 
> watch file formats.

It may be worth suggesting that to the Lintian developers

>> debian/changelog
>>  - Most new packages use a single entry in the changelog, for the
>> initial packaging and closing of ITP bug
> 
> That makes sense, but OTOH, I’m using these packages internally in my 
> personal work.  So as I find bugs I am bumping the version internally … it 
> makes sense to me to keep meaningful changelog entries.  Is there a reason 
> they’re considered harmful, or is it just a matter of personal preference?  I 
> suspect you are right that I should push the “Closes: #738850” entry to the 
> release that actually closes the ITP … I’ll do that when I go to submit.

I don't think it's necessarily considered bad, but something I've
noticed a lot of Debian Developers require when they sponsor packages on
mentors.debian.org

>> I would be reluctant to take over the package entirely as I'm not really
>> familiar with library packaging (other than perl modules).
> 
> This one, thankfully, is a pretty easy one.  No pressure was intended … I’m 
> happy to maintain it.  Just wanted to extend the offer.

No problem ;)

Regards
-- 
Daniel Lintott
GPG Key: 4096R/5D73EC6E

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to