1 |
On 12/16/19 12:27 AM, Michał Górny wrote: |
2 |
> On Sun, 2019-12-15 at 13:29 -0800, Zac Medico wrote: |
3 |
>> On 12/13/19 2:12 PM, Michael 'veremitz' Everitt wrote: |
4 |
>>> On 13/12/19 20:36, Michał Górny wrote [excerpted]: |
5 |
>>>> Is this really an argument for or *against* it? Developers are entirely |
6 |
>>>> capable of keeping seds that do nothing for years, as well as patches |
7 |
>>>> that -- while apparently applying correctly -- are entirely meaningless. |
8 |
>>> <snip> |
9 |
>>> |
10 |
>>> I think there is some merit in some kind of feedback when sed's are doing |
11 |
>>> nothing, although how feasible it is to generate any useful feedback I |
12 |
>>> can't say. I wouldn't say it needs to explicitly fail or make lots of |
13 |
>>> noise, just an info message that could prompt some further investigation. |
14 |
>>> |
15 |
>> |
16 |
>> It's possible to implement a sed wrapper that detects file arguments for |
17 |
>> -i/--in-place mode, and compares file content before and after the sed call. |
18 |
>> |
19 |
>> There are also ways to make sed exit with an error but that won't be as |
20 |
>> easy to use as a sed wrapper: |
21 |
>> |
22 |
>> https://stackoverflow.com/questions/15965073/return-code-of-sed-for-no-match/15966279 |
23 |
> |
24 |
> Don't forget that there could be valid cases for sed not changing |
25 |
> a file. Not to mention corner cases where a working replacement results |
26 |
> in no change, e.g.: |
27 |
> |
28 |
> # yuck! |
29 |
> sed -i -e "s^-O2^${CFLAGS}^" ... |
30 |
> |
31 |
|
32 |
Still, I imagine the number of false positives for a QA warning based on |
33 |
this would be relatively low. |
34 |
-- |
35 |
Thanks, |
36 |
Zac |