| 1 |
On Fri, Apr 24, 2020 at 05:07:48PM -0400, Michael Orlitzky wrote: |
| 2 |
> I might believe you about speed, but not about RAM. Memory usage goes up |
| 3 |
> with static linking because you've got multiple copies of the same thing |
| 4 |
> loaded into memory. |
| 5 |
|
| 6 |
No. I told about RAM :-). Several years ago i had some research for one |
| 7 |
project. It needed to run multiple instances of the same program(several |
| 8 |
thousands of concurrent instances). We tried to achieve maximum memory |
| 9 |
economy. And we saw that when the program linked statically, each instance |
| 10 |
consume less memory starting from 6 instances. Thanks to sharing of .text |
| 11 |
segments. |
| 12 |
|
| 13 |
Thus, for something like bash a static linking isn't bad. I have now 12 |
| 14 |
instances of it running. If it would be static, then not only every script |
| 15 |
that i run during work day starts faster, but it consume a little less ram. |
| 16 |
|
| 17 |
> think I'm wrong, feel free to shoot yourself in the foot, but you |
| 18 |
> shouldn't be calling Alessandro or the QA team incompetent (that's my |
| 19 |
> bit...) unless you have some strong new evidence that static linking |
| 20 |
> improves things in a general-purpose linux distro. |
| 21 |
|
| 22 |
No-no. I didn't want to call QA team or Alessandro incompetent. May |
| 23 |
be some typo or misspelling. I just said that anybody who says "Nothing should |
| 24 |
be statically linked" is incompetent in this question. Because this is simply |
| 25 |
not true. |
| 26 |
|
| 27 |
|
| 28 |
-- |
| 29 |
Олег Неманов (Oleg Nemanov) |
Archives