| 1 |
On 06/07/2016 12:20 PM, Michał Górny wrote: |
| 2 |
>> |
| 3 |
>> A pkg_pretend() message would certainly make sense and IMO be a good |
| 4 |
>> idea, but again this isn't any different than the situation as it |
| 5 |
>> stands now WITHOUT a USE=gui. Regardless I don't see this as a |
| 6 |
>> blocker to the idea. |
| 7 |
> |
| 8 |
> Nope, it won't. It will only make things slower, more clumsy |
| 9 |
> and provide no real benefit. pkg_pretend() should be reserved for |
| 10 |
> checks really needed to build packages and not abused to do some early |
| 11 |
> output. |
| 12 |
> |
| 13 |
|
| 14 |
I forgot to reply to this. |
| 15 |
|
| 16 |
The benefit of pkg_pretend is that it lets you give a sensible error |
| 17 |
message. This is not a complex problem we're asking the user to resolve. |
| 18 |
Here's what the error would say if I used pkg_pretend: |
| 19 |
|
| 20 |
This package supports two different toolkits, gtk2 and gtk3. Please |
| 21 |
choose one and add it to your USE flags for <pkg>. |
| 22 |
|
| 23 |
Here's what it says with REQUIRED_USE: |
| 24 |
|
| 25 |
!!! Problem resolving dependencies for x11-drivers/nvidia-drivers |
| 26 |
from @selected |
| 27 |
... done! |
| 28 |
|
| 29 |
!!! The ebuild selected to satisfy "x11-drivers/nvidia-drivers" has |
| 30 |
unmet requirements. |
| 31 |
- x11-drivers/nvidia-drivers-352.30::gentoo USE="X multilib tools |
| 32 |
-acpi -gtk2 -gtk3 -pax_kernel -uvm" ABI_X86="64" |
| 33 |
|
| 34 |
The following REQUIRED_USE flag constraints are unsatisfied: |
| 35 |
tools? ( any-of ( gtk2 gtk3 ) ) |
| 36 |
|
| 37 |
The above constraints are a subset of the following complete |
| 38 |
expression: |
| 39 |
tools? ( X any-of ( gtk2 gtk3 ) ) |
| 40 |
|
| 41 |
We could make the portage output a little better, but that's going to be |
| 42 |
an uphill battle and still won't be as informative or short as the |
| 43 |
pkg_pretend version. |
| 44 |
|
| 45 |
What we really need is a way to tie an error message to a REQUIRED_USE |
| 46 |
clause. I think I remember Ciaran sarcastically suggesting that we make |
| 47 |
REQUIRED_USE a dictionary lookup from clause => message. The sarcasm was |
| 48 |
probably because all we'd be doing is replacing the if/then/die |
| 49 |
statements (that we'd use in pkg_pretend) with ?,(), and => in |
| 50 |
REQUIRED_USE. It's an abstraction without any benefit at that point. |
Archives