two questions on file modes
Shiro Kawai
(24 Jul 2020 20:15 UTC)
|
Re: two questions on file modes
hga@xxxxxx
(24 Jul 2020 21:16 UTC)
|
Re: two questions on file modes
Shiro Kawai
(24 Jul 2020 21:48 UTC)
|
Re: two questions on file modes
Lassi Kortela
(26 Jul 2020 19:24 UTC)
|
Re: two questions on file modes
Lassi Kortela
(26 Jul 2020 19:39 UTC)
|
Re: two questions on file modes
hga@xxxxxx
(26 Jul 2020 20:18 UTC)
|
Re: two questions on file modes
Lassi Kortela
(26 Jul 2020 20:33 UTC)
|
Re: two questions on file modes
John Cowan
(26 Jul 2020 23:15 UTC)
|
Re: two questions on file modes Lassi Kortela (27 Jul 2020 06:09 UTC)
|
Re: two questions on file modes
John Cowan
(28 Jul 2020 20:31 UTC)
|
> SRFI 170, however, is designed to provide a minimum of either fd or port > operations. > You can translate between one and the other, but the > intention is to use ports internally and fds only externally. That makes sense when a mismatch between ports and fd's is in question, i.e. the fd has some property that should be mirrored in the port object, and a discrepancy can develop between the fd's and the port object's idea of that property. However, some operations merely act on the underlying file. Then I don't think it matters very much, since Scheme won't get out of sync and being able to pass either a port or an fd is convenient. > IMO, if you have a stale pathname, you are most likely doing the Wrong > Thing anyway. It's a surprisingly common problem in long-running programs. For example, a shell should cope with the situation where the pathname for the current directory disappears from underneath. Programs are not always in control of doing the right thing since another unrelated program can mess things up.