[pacman-dev] [PATCH] Adds a "retry?" prompt for when transaction creation fails for user-friendlier fail support.

[Eli Schwartz]

On 4/28/20 8:40 PM, Allan McRae wrote:
> On 28/4/20 9:43 am, Pedro wrote:
>> EDIT: Rewrote the prompt message and added a check for the --noconfirm flag
>>
>> >From what I could see, the function trans_init that comes right before the prompt has already specific messages
>> for each operation, so I consider that this new implementation fits the purpose.
>> My motivation to submit this patch is that it's pretty common for AUR-helpers to automatically invoke pacman after a build,
>> but correct transaction error handling is quite rare. So I thought that to implement a handling mechanism in the user side
>> would solve the problem not only in that case, but also for other scripts that invoke pacman.
>>
>> Signed-off-by: Pedro <pedrocga1 at gmail.com>
>> ---
> 
> I don't see how retrying the same operation with the same input is going
> to help anyone.  If pacman fails in these ways, it is not going to
> succeed the next attempt.
> 
> Allan
Just like I said the first time this patch was submitted -- and now it
just has a slightly longer "retry" message, but it still doesn't answer
my question "what is the user supposed to do about it"?

Pedro, a AUR helper is most likely going to fail because of dependency
checks, which happen later. Do you have examples of a situation where
pacman cannot even create the transaction, let alone populate it with
packages? If so, why is it happening?

These are questions that should be answered before adding a prompt with
no clear purpose.

-- 
Eli Schwartz
Bug Wrangler and Trusted User

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 1601 bytes
Desc: OpenPGP digital signature
URL: <https://lists.archlinux.org/pipermail/pacman-dev/attachments/20200428/f816e2d3/attachment.sig>