New /etc/services for testing available

Bill Crawford billcrawford1970 at gmail.com
Wed Mar 8 16:53:45 UTC 2006


Rahul Sundaram wrote:
> Bill Crawford wrote:
>
>>
>> I'd really suggest instead making a poll of what people would find 
>> useful.
>>
>> There are going to be a lot of things in there that will probably 
>> never be useful, e.g. 2545 "sis-emt" is one I registered, and I'm 
>> pretty sure is now unused, or certainly of only very limited use.
>>
>> Just a thought.
>
> Whats the harm in having them there though?
>
Wasted time, mostly. I know we're talking about less than a second per, 
but it's still unnecessary.

Harm would mostly be e.g. confusion caused to people looking at netstat 
output, though I'm pleased to see things like the AOL port numbers 
listed there (which is certainly common enough that I see it primarily 
useful; however even then .. *shrug*).

Anyway, it's wasted time for an app to search. Granted I'm only 
referring to one particular entry, but I can guarantee that for the 
majority of people out there, if a connection happens to be on port 
2545, having it show up as "sis-emt" isn't helping them one bit :)

Same probably goes for a lot of the others, is the point I was making. 
If that ends up being a majority of them (which I suspect) it really is 
just wasted time. Trivial in many cases, yes, but still quite 
unnecessary, and again, probably confusing more than helpful anyway. In 
other words:-

Cost: time.
Benefit: loss of time figuring out why a connection shows up as "foo" in 
netstat output.

(i.e. the "benefit" side is negative too.)

FWIW I do agree that having a few more than the barest minimum is 
probably useful. But multiplying up the time taken to scan /etc/services 
significantly for no or possibly negative benefit is just (IMO) poor 
engineering choice.

Again, I'm explicitly talking about a subset of what's there, not saying 
"don't expand it, DANGER WILL ROBINSON DANGER" without thought. I *know* 
at least one entry in there is pretty useless. And I'm sure there will 
be others.

I'm asking for a positive input of what folk feel *should* be in there, 
and suggesting that all such be included. Hoping it'll be less than the 
full list :)




More information about the fedora-devel-list mailing list