|Subject:||Re: [Liberty-eiffel] PATH_NAME. make from string|
|Date:||Thu, 24 Jul 2014 10:53:36 +0200|
On 24 luglio 2014 00:27:09 CEST, Cyril ADRIAN <address@hidden> wrote:
>2014-07-23 23:55 GMT+02:00 Paolo Redaelli <address@hidden>:
>> Just a quick request:
>> Would you mind if I change PATH_NAME.make_from_string to accept
>I would not mind :-)
>... as long as the implementation stays very memory conservative.
Being very memory conservative is a paramount design requirement in standard libraries.
I was naively going to write that turning all *input* STRINGs in standard library into ABSTRACT_STRINGs shouldn't have changed the memory usage when the input is actually a STRING. It wouldn't since most in places aren't easy with the seemingly unmutable ABSTRACT_STRING.
Thinking better about it it may be wiser to use ABSTRACT_STRING.string feature in user code. I would remove the "put for elks compatibility" note: is the old elks relevant anymore? I don't think so. "reasonable, best effort" compatibility with ISE is somehow preferable, for a feasible and comfortable definition on the terms :-)
>> A mostly offline Paolo.... :-)
>Have nice holidays! :-)
Thanks... Almost finished, as Saturday we will be on the back journey.
I discovered being a bandwidth vampire: I "smoked" more than 700mb of data in ten days when I tried to be conservative... :-)
>Cyril, in holidays too, but working on backup automation procedures at
>(hence the late hour) :-)
That's reminds me that I shall ditch GlusterFS in favor of Ceph AND deploy a "real" backup. What's your approach? Offline, online, remote, on the cloud ?
|[Prev in Thread]||Current Thread||[Next in Thread]|