kicad-developers team mailing list archive

[Dick Hollenbeck <dick@xxxxxxxxxxx>]

On 07/28/2010 02:18 PM, Wayne Stambaugh wrote:
> On 7/28/2010 6:03 AM, Dick Hollenbeck wrote:
>   
>> On 07/21/2010 04:31 PM, Dick Hollenbeck wrote:
>>     
>>> Are people happy with launchpad?
>>>
>>> 10 = extremely happy
>>> :
>>> :
>>> 1 = extremely dissatisfied
>>>    
>>>       
>>
>> Thanks to those that responded.  I am pleased that everyone is mostly pleased. 
>> This gives me political cover to now offer my own opinion.
>>
>> Here are my ratings:
>>
>>
>> Bugtracker:  3
>>
>> Patch handling:  1
>>
>> Wiki or generic webpage infrastructure: 1
>>
>> Bzr: 8  (I like "bzr qlog")
>>
>> Maillist: 7
>>
>> Launchpad.net overall: 3
>>
>>
>> The world of open source hosting has a long way to go in my opinion.
>>
>>
>> My biggest surprise, unexpected disappointment and complaint with launchpad is
>> the patch handling.  This is compounded by the fact that it was a (almost the)
>> primary motivator for the move TO launchpad, IMO.
>>
>>
>> Let me summarize my frustration and ask for some help on a path forward:
>>
>>
>> There are two pathways into the "Active Reviews"
>> (https://code.launchpad.net/kicad/+activereviews) list, which is the so called
>> patch handler:
>>
>>
>> 1) Push a modified branch to launchpad, "Propose" a merge request at the
>> launchpad.net website,
>> https://code.launchpad.net/~kicad-testing-committers/kicad/testing/+register-merge
>>
>>
>> 2) Locally, $ bzr send, and send the "merge bundle" as an attachment to a
>> signed email.
>>
>>
>>
>> The problem with 1) is that it requires too much learning for a "drive by patch
>> submitter" person.  The learning investment is too high for somebody that would
>> submit about 5 patches per year or less.  We will still end up with patches on
>> the mailing list.
>>
>>
>> The problem with 2) is that the email parser at launchpad.net will not handle
>> any variations other than some undocumented concoction of signed main mail body
>> but with an unsigned bundle attachment.  I am unable to get this to work after
>> 3 days trying.  Help on irc #launchpad usually comes back saying "it works in
>> the general case", wait for so and so to help you.  So and so comes later and
>> says the same thing.  I have worn out my political capital there.  They know I
>> am extremely unhappy and have told me they don't care if I were to simply go
>> away, along with the whole Kicad project.
>>
>>
>> So, what is a merge bundle?  A merge bundle is basically a patch generated by
>> BZR, with special meta data at the end which holds the name of the destination
>> branch, submitter, author, etc.
>>
>>
>> And why is using a signed email the only way to submit a merge bundle?  Who the
>> hell knows.
>>     
> Dick,
>
> Apparently I have run into some of the problems you did with option 2
> attempting to send a test merge request.  I am currently waiting for a reply
> for error I got back from Launchpad so I'll continue to attempt to make some
> progress once I get a response.  In any event, you are correct that it is a
> PITA to get your mail client (in my case Thunderbird) setup and all of the GPG
> key stuff set up just to submit a merge request.  I was curious how other
> Launchpad projects were handling this issue.  I checked about 5 or 6 random
> projects and could not find a single instance of a merge request that was not
> submitted against a public branch pushed to Launchpad (option 1).  Maybe there
> is a reason no one uses option 2.
>
>   
>> I ask you to help me get the launchpad developers to accept a merge bundle
>> through a web form.
>>
>> There is no reason that if a merge bundle is good enough input in general, that
>> a merge bundle coming in through another secure pathway is not also acceptable.
>>     
> I don't understand why there isn't already a web interface to support merge
> requests in this manner.  Once you are logged into Launchpad, all of you
> changes go through https so I don't see an issue.  It would still be nice to be
> able to submit merge requests directly from you mail client.  Firing up a
> browser and logging in to submit a merge request are extra steps that can be
> avoided.  I agree that for folks who submit simple infrequent patches should
> not have to go through the complication of options 1 and 2.
>
>   
>> I am asking for help in getting some action going on it by the developers
>> through https://answers.launchpad.net/launchpad
>>     
> Here is the link to the question I had from the error response from launchpad:
>
> https://answers.edge.launchpad.net/launchpad-code/+question/119254
>
> I will continue to follow up when I get an answer and/or a solution.  If the
> answer and/or solution I get is unsatisfactory, I will submit a request to add
> a form to the merge request page for uploading merge bundles.
>
> Wayne
>   

Thanks Wayne. I got further than you have with dialog to the developers
using IRC so it was not recorded fully.  The end result was "sorry, the
logging of the error message is insufficient in the current code".  They
have an error log, but it is not descriptive enough.


If I ask for the new feature, it will not carry as much weight as if
somebody else asks for it.  Remember, I have used up all my political
capital with these mere launchpad developers. 


This was the *main purpose* of this thread, me asking for help.    Well
technically, this is me asking for help in order that I might better
help others, including those helping me.

I don't think we should delay to ask for the feature.  Submitting a
patch (merge bundle) at a webform is easy as cake.  They have half the
code already written.

Why limit the number of people making the request to one person?  Also,
I see no reason why a personal email could not be sent to Mark
Shuttleworth, who is listed as a project administrator of launchpad, and
IIRC is the owner of Canonical.  I suspect if he new all his work was
getting a score of 3 from a launchpad.net user (me), it might mean
something.  If not, then my score of 3 gets lowered.


Thanks Wayne.

Dick