Closed Bug 71004 Opened 24 years ago Closed 12 years ago

Proxy: "HTTP" and "FTP" used before "SOCKS"

Categories

(Core :: Networking, defect)

All
Windows NT
defect
Not set
normal

Tracking

()

RESOLVED WONTFIX

People

(Reporter: venur, Unassigned)

References

()

Details

(Keywords: helpwanted)

From Bugzilla Helper: User-Agent: Mozilla/4.73 [en] (WinNT; I) BuildID: 2001022805 If we give any proxy address for socks proxy, it should first verify for it in the DNS as N4.X does. Reproducible: Always Steps to Reproduce: 1.Open Preferences window 2.Select Manual Proxy configuration 3.give valid proxies for htt pproxy and ftp proxy 4.give some junk proxy address for Socks proxy and port no :1080 5.Say OK on preferences dialog and try to access any web site. Actual Results: It is not showing any alert, it just connects to the website through HTTP proxy. Expected Results: it should show an alert after validating the proxies given that the Socks proxy is not a valid proxy or doesn't have DNS entry
so is this a socks issue, does it belong to the nonexistant proxies category, or is it really a preferences bug?
->networking
Assignee: matt → neeti
Component: Preferences → Networking
QA Contact: sairuh → tever
My understanding is that this is the correct behavior for Communicator 4. http://www.packetgram.com/pktg/docs/nscp/ns6/ns6proxy.html#comm4users I do not like this UI, it needs to be redesigned so order of server usage is clear.
Summary: Socks proxy must be verified for DNS entry.. → SOCKS - "HTTP Proxy" and "FTP Proxy" settings used first.
Marking NEW.
Status: UNCONFIRMED → NEW
Ever confirmed: true
Summary: SOCKS - "HTTP Proxy" and "FTP Proxy" settings used first. → SOCKS - "HTTP Proxy" and "FTP Proxy" settings used first instead of warning about DNS issue
QA to me.
QA Contact: tever → benc
Target Milestone: --- → mozilla1.0
So what is this bug really asking for? Is there a problem with the priority of the proxies used? Or a problem with the UI? Or should we try to access the proxies when the prefs dialog is closed, rejecting those which can't be accessed at the time the preference is set? I don't like rejecting the server at the preferences dialog if it can't be reached at the moment.
We need to know the 4xp behavior first.
Keywords: 4xp
Bugs targeted at mozilla1.0 without the mozilla1.0 keyword moved to mozilla1.0.1 (you can query for this string to delete spam or retrieve the list of bugs I've moved)
Target Milestone: mozilla1.0 → mozilla1.0.1
Target Milestone: mozilla1.0.1 → Future
If an HTTP proxy is set, and you are using the HTTP protocol for a connection, it uses the HTTP proxy rather than the SOCKS proxy. Mozilla uses SOCKS when nothing else is applicable. If SOCKS were chosen over HTTP proxies, and the SOCKS server entry was invalid, it would not connect. It would behave as it does when only an invalid SOCKS proxy was set. That is, it says it cannot connect to whatever host you were trying to go to. It doesn't give you a warning about the invalid proxy server, which is bad but a slightly different issue. A bad HTTP proxy entry gives a different, but equally bad error, too. This is the relevant code for the proxy order: http://lxr.mozilla.org/seamonkey/source/netwerk/base/src/nsProtocolProxyService.cpp#332 I think this bug should be marked invalid, and a new bug opened regarding the lack of bad proxy error messages. Additionally, we might want to consider another bug regarding the order of proxy selections. Personally, I think the current order makes sense (I actually chose this particular behavior when I first added SOCKS), as there might be advantages to having clients go through protocol specific proxies before generic ones. A company could more easily log web usage through an HTTP proxy, for example. Or it could use a caching HTTP proxy, or strip cookies, etc.
moving neeti's futured bugs for triaging.
Assignee: neeti → new-network-bugs
This bug should remain open because it is a 4xp bug. Communicator used SOCKS first, then applications proxies second. Mozilla does the opposite, which could create problems for users if they use profile migration. That is a point I failed to note in the packetgram.com document above, I've made a correction. The proxy error needs to be fixed in bug 125445. As for fixing the order of proxy usage, or adding failover, etc, the only feasible solution I can think of is bug 89928, where we replace the fixed solutions we have w/ a more flexible PAC based architecture.
Summary: SOCKS - "HTTP Proxy" and "FTP Proxy" settings used first instead of warning about DNS issue → Proxy: "HTTP" and "FTP" used before "SOCKS"
If this bug is just a matter of using SOCKS before HTTP and FTP proxies, then it's very easy to fix. So do we want to do this? I still believe the current behavior is better, 4xp or not.
Protocol specific before socks is definately the correct behavior, IMO. Consider that I need socks for my mail/news but still want to use a caching http proxy for speed (or ad filtering, or tracking, or whatever). I suppose you could create a more specific UI so that I would be able to specify what I want to use socks for, but as it is the more generic protocol should be overridden by the others, despite what NS did.
Darin, this was reported by our india test team
-> me
Assignee: general → darin
Target Milestone: Future → mozilla1.8alpha
Status: NEW → ASSIGNED
OS: Windows NT → Windows 3.1
Hardware: PC → All
Darin: from a design standpoint, I guess the first thing you should do is decide which behavior is better, from a behavioral or "4xp" standpoint. I'm not sure I have an opinion anymore, anyone trying to use both application and connection-level proxies together should be using PAC.
OS: Windows 3.1 → Windows NT
Keywords: helpwanted
Target Milestone: mozilla1.8alpha1 → Future
Assignee: darin → nobody
Status: ASSIGNED → NEW
QA Contact: benc → networking
Target Milestone: Future → ---
to the extent this is a profile migration compat problem from ns 4.. wontfix
Status: NEW → RESOLVED
Closed: 12 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.