[pacman-dev] [PATCH] libalpm: fix alpm_fetch_pkgurl with a fetch callback

[Guillaume Benoit]

Le 09/05/2021 à 15:18, Allan McRae a écrit :
> On 5/5/21 11:54 pm, Guillaume Benoit wrote:
>> Le 04/05/2021 à 01:30, Allan McRae a écrit :
>>> On 3/5/21 9:28 pm, Guillaume Benoit wrote:
>>>>
>>>> If I have time to work on another version of a patch that pass the full
>>>> payload to
>>>> the front-end, is there a chance that it could be included in pacman
>>>> 6.0 ?
>>>
>>> There is a chance...
>>>
>>> The things going against me accepting it for 6.0 are I have called a
>>> freeze apart from already submitted patches and regressions.  This is
>>> not a regression.  I guess there will also be an API change.
>>>
>>> However, I am slow at reviewing the last patches I need to apply, so a
>>> patch that appears soon and I judge as having minimal risk for the
>>> internal download may be accepted.
>>>
>>> Allan
>>>
>>
>> v2: I choosed to create another alpm_download_payload struct to only expose
>> required fields to the API, alpm_cb_fetch callback now has this struct
>> as argument.
>> Those are the only API changes.
>> I also rewrote the download_with_xfercommand function in pacman code.
>> What is fixed:
>> - download from an url with a fetch callback for any front-end
>> - download from an standard url with pacman with a xfercommand
>> What is not fixed:
>> - download from an url, with pacman with an xfercommand, when this url
>> doesn't
>> contain the filename like
>> https://archlinux.org/packages/core/x86_64/pacman/download/
>>
> 
> Does this also replace:
> libalpm: download signatures with a fetch callback
> 
> That was on my list next, but I took a quick skim at this patch and it
> appears to supersede that patch.
> 
> Allan
> 
Yes, with this patch, signatures download is under the responsibility of the fetch callback,
checking payload->download_signature.
That's now implemented in pacman's download_with_xfercommand function.