On 30.05.2011 09:45, Florian Fainelli wrote:
> On Sunday 29 May 2011 20:18:46 Philip Prindeville wrote:
>> On 5/29/11 10:34 AM, John Crispin wrote:
>>> On 29/05/11 19:14, Philip Prindeville wrote:
>>>> Minor comment. It's my understanding that gmake will fork/exec commands
> directly if it doesn't detect any meta-characters, thereby saving a fork/exec
> pair of $(SHELL) to interpret them. In this case:
>>>>> + $(CP) $(PKG_INSTALL_DIR)/usr/lib/ipsec/{openac,pki,scepclient}
>>>>> $(1)/usr/lib/ipsec/
>>>>
>>>> it would be a lot more efficient to just write out both pathnames as
>>>> absolute (easier to grep through logs, too).
>>>>
>>>> -Philip
>>>
>>> i disagree, the proposed call is consistent with how most other owrt
>>> Makefiles are written
>>
>> I'm not saying it doesn't match what's been done elsewhere: I'm just saying
>> it's not the most efficient way to do things, which is pretty
>> incontrovertible.
>
> Do we really care about performance at package install time? Certainly not,
> so
> let's keep it that way, and encourage people writing things that way as well,
> because it makes the Makefile shorter and concise.
I even tend to use wildcards or find just to have not change these install
instructions everytime a new binary is added upstream. But ntfs-3g is kinda
special as it by now is split into three virtual packages
(ntfs-3g,-utils,ntfstools_).
bud
_______________________________________________
openwrt-devel mailing list
openwrt-devel@lists.openwrt.org
https://lists.openwrt.org/mailman/listinfo/openwrt-devel
