cURL and libcurl, Web

Easier HTTP requests with h2c

4 Comments

I spend a large portion of my days answering questions and helping people use curl and libcurl. With more than 200 command line options it certainly isn’t always easy to find the correct ones, in combination with the Internet and protocols being pretty complicated things at times… not to mention the constant problem of bad advice. Like code samples on stackoverflow that repeats non-recommended patterns.

The notorious -X abuse is a classic example, or why not the widespread disease called too much use of the –insecure option (at a recent count, there were more than 118,000 instances of “curl --insecure” uses in code hosted by github alone).

Sending HTTP requests with curl

HTTP (and HTTPS) is by far the most used protocol out of the ones curl supports. curl can be used to issue just about any HTTP request you can think of, even if it isn’t always immediately obvious exactly how to do it.

h2c to the rescue!

h2c is a new command line tool and associated web service, that when passed a complete HTTP request dump, converts that into a corresponding curl command line. When that curl command line is then run, it will generate exactly(*) the HTTP request you gave h2c.

h2c stands for “headers to curl”.

Many times you’ll read documentation somewhere online or find a protocol/API description showing off a full HTTP request. “This is what the request should look like. Now send it.” That is one use case h2c can help out with.

Example use

Here we have an HTTP request that does Basic authentication with the POST method and a small request body. Do you know how to tell curl to send it?

The request:

POST /receiver.cgi HTTP/1.1
Host: example.com
Authorization: Basic aGVsbG86eW91Zm9vbA==
Accept: */*
Content-Length: 5
Content-Type: application/x-www-form-urlencoded

hello

I save the request above in a text file called ‘request.txt’ and ask h2c to give the corresponding curl command line:

$ ./h2c < request.txt
curl --http1.1 --header User-Agent: --user "hello:youfool" --data-binary "hello" https://example.com/receiver.cgi

If we add "--trace-ascii dump” to that command line, run it, and then inspect the dump file after curl has completed, we can see that it did indeed issue the HTTP request we asked for!

Web Site

Maybe you don’t want to install another command line tool written by me in your system. The solution is the online version of h2c, which is hosted on a separate portion of the official curl web site:

https://curl.se/h2c/

The web site lets you paste a full HTTP request into a text form and the page then shows the corresponding curl command line for that request.

h2c “as a service”

Inception alert: you can also use the web version of h2c by sending over a HTTP request to it using curl. You’ll then get nothing but the correct curl command line output on stdout.

To send off the same file we used above:

curl --data-urlencode http@request.txt https://curl.se/h2c/

or of course if you rather want to pass your HTTP request to curl on stdin, that’s equally easy:

cat request.txt | curl --data-urlencode http@- https://curl.se/h2c/

Early days, you can help!

h2c was created just a few days ago. I’m sure there are bugs, issues and quirks to iron out. You can help! Files issues or submit pull-requests!

(*) = barring bugs, there are still some edge cases where the exact HTTP request won’t be possible to repeat, but where we instead will attempt to do “the right thing”.

cURL and libcurl, Open Source, Security

keep finding old security problems

I decided to look closer at security problems and the age of the reported issues in the curl project.

One theory I had when I started to collect this data, was that we actually get security problems reported earlier and earlier over time. That bugs would be around in public release for shorter periods of time nowadays than what they did in the past.

My thinking would go like this: Logically, bugs that have been around for a long time have had a long time to get caught. The more eyes we’ve had on the code, the fewer old bugs should be left and going forward we should more often catch more recently added bugs.

The time from a bug’s introduction into the code until the day we get a security report about it, should logically decrease over time.

What if it doesn’t?

First, let’s take a look at the data at hand. In the curl project we have so far reported in total 68 security problems over the project’s life time. The first 4 were not recorded correctly so I’ll discard them from my data here, leaving 64 issues to check out.

The graph below shows the time distribution. The all time leader so far is the issue reported to us on March 10 this year (2017), which was present in the code since the version 6.5 release done on March 13 2000. 6,206 days, just three days away from 17 whole years.

There are no less than twelve additional issues that lingered from more than 5,000 days until reported. Only 20 (31%) of the reported issues had been public for less than 1,000 days. The fastest report was reported on the release day: 0 days.

The median time from release to report is a whopping 2541 days.

When we receive a report about a security problem, we want the issue fixed, responsibly announced to the world and ship a new release where the problem is gone. The median time to go through this procedure is 26.5 days, and the distribution looks like this:

What stands out here is the TLS session resumption bypass, which happened because we struggled with understanding it and how to address it properly. Otherwise the numbers look all reasonable to me as we typically do releases at least once every 8 weeks. We rarely ship a release with a known security issue outstanding.

Why are very old issues still found?

I think partly because the tools are gradually improving that aid people these days to find things much better, things that simply wasn’t found very often before. With new tools we can find problems that have been around for a long time.

Every year, the age of the oldest parts of the code get one year older. So the older the project gets, the older bugs can be found, while in the early days there was a smaller share of the code that was really old (if any at all).

What if we instead count age as a percentage of the project’s life time? Using this formula, a bug found at day 100 that was added at day 50 would be 50% but if it was added at day 80 it would be 20%. Maybe this would show a graph where the bars are shrinking over time?

But no. In fact it shows 17 (27%) of them having been present during 80% or more of the project’s life time! The median issue had been in there during 49% of the project’s life time!

It does however make another issue the worst offender, as one of the issues had been around during 91% of the project’s life time.

This counts on March 20 1998 being the birth day. Of course we got no reports the first few years since we basically had no users then!

Specific or generic?

Is this pattern something that is specific for the curl project or can we find it in other projects too? I don’t know. I have not seen this kind of data being presented by others and I don’t have the same insight on such details of projects with an enough amount of issues to be interesting.

What can we do to make the bars shrink?

Well, if there are old bugs left to find they won’t shrink, because for every such old security issue that’s still left there will be a tall bar. Hopefully though, by doing more tests, using more tools regularly (fuzzers, analyzers etc) and with more eyeballs on the code, we should iron out our security issues over time. Logically that should lead to a project where newly added security problems are detected sooner rather than later. We just don’t seem to be at that point yet…

Caveat

One fact that skews the numbers is that we are much more likely to record issues as security related these days. A decade ago when we got a report about a segfault or something we would often just consider it bad code and fix it, and neither us maintainers nor the reporter would think much about the potential security impact.

These days we’re at the other end of the spectrum where we people are much faster to jumping to a security issue suspicion or conclusion. Today people report bugs as security issues to a much higher degree than they did in the past. This is basically a good thing though, even if it makes it harder to draw conclusions over time.

Data sources

When you want to repeat the above graphs and verify my numbers:

  • vuln.pm – from the curl web site repository holds security issue meta data
  • releaselog – on the curl web site offers release meta data, even as a CSV download on the bottom of the page
  • report2release.pl – the perl script I used to calculate the report until release periods.
cURL and libcurl

Some things to enjoy in curl 7.55.0

In this endless stream of frequent releases, the next release isn’t terribly different from the previous.

curl’s 167th release is called 7.55.0 and while the name or number isn’t standing out in any particular way, I believe this release has a few extra bells and whistles that makes it stand out a little from the regular curl releases, feature wise. Hopefully this will turn out to be a release that becomes the new “you should at least upgrade to this version” in the coming months and years.

Here are six things in this release I consider worthy some special attention. (The full changelog.)

1. Headers from file

The command line options that allows users to pass on custom headers can now read a set of headers from a given file.

2. Binary output prevention

Invoke curl on the command line, give it a URL to a binary file and see it destroy your terminal by sending all that gunk to the terminal? No more.

3. Target independent headers

You want to build applications that use libcurl and build for different architectures, such as 32 bit and 64 bit builds, using the same installed set of libcurl headers? Didn’t use to be possible. Now it is.

4. OPTIONS * support!

Among HTTP requests, this is a rare beast. Starting now, you can tell curl to send such requests.

5. HTTP proxy use cleanup

Asking curl to use a HTTP proxy while doing a non-HTTP protocol would often behave in unpredictable ways since it wouldn’t do CONNECT requests unless you added an extra instruction. Now libcurl will assume CONNECT operations for all protocols over an HTTP proxy unless you use HTTP or FTP.

6. Coverage counter

The configure script now supports the option –enable-code-coverage. We now build all commits done on github with it enabled, run a bunch of tests and measure the test coverage data it produces. How large share of our source code that is exercised by our tests. We push all coverage data to coveralls.io.

That’s a blunt tool, but it could help us identify parts of the project that we don’t test well enough. Right now it says we have a 75% coverage. While not totally bad, it’s not very impressive either.

Stats

This release ships 56 days since the previous one. Exactly 8 weeks, right on schedule. 207 commits.

This release contains 114 listed bug-fixes, including three security advisories. We list 7 “changes” done (new features basically).

We got help from 41 individual contributors who helped making this single release. Out of this bunch, 20 persons were new contributors and 24 authored patches.

283 files in the git repository were modified for this release. 51 files in the documentation tree were updated, and in the library 78 files were changed: 1032 lines inserted and 1007 lines deleted. 24 test cases were added or modified.

The top 5 commit authors in this release are:

  1. Daniel Stenberg
  2. Marcel Raad
  3. Jay Satiro
  4. Max Dymond
  5. Kamil Dudka
cURL and libcurl, Open Source

The curl bus factor

4 Comments

bus factor: the minimum number of team members that have to suddenly disappear from a project before the project stalls due to lack of knowledgeable or competent personnel.

Projects should strive to survive

If a project is worth using and deploying today and if it is a project worth sending patches to right now, it is also a project that should position itself to survive a loss of key individuals. However unlikely or unfortunate such an event would be.

Tools to calculate bus factor

All the available tools that determine the bus factor for a given project only run on code and check for commits, code churn or check how many files each person has done a significant share of changes in etc.

This number is really impossible to figure out without tools and tools really cannot take “general knowledge” into account, or “this person answers a lot of email on the list”, or this person has 48k in reputation on stack overflow already for responding to questions about the project.

The bus factor as evaluated by a tool pretty much has to be about amount of code, size of code or number of code changes, which may or may not be a good indicator of who knows what about the code. Those who author and commit changes probably have a good idea but a real problem is that you can’t reverse that view and say that just because you didn’t commit or change something, you don’t know. Do you know more about the code if you did many commits? Do you know more about the code if you changed more lines of code?

We can’t prove or assume lack of knowledge or interest by an absence of commits, edits or changes. And yet we can’t calculate bus factor if there’s no tool or way to calculate it.

A look at curl

curl is soon 20 years old and boasts 22k something commits. I’m the author of about 57% of them, and the second-most committer (who’s not involved anymore) has about 12%. That makes two committers having done 15.3k commits out of the 22k. If we for simplicity calculate bus factor based on commit numbers, we’d need 8580 commits from others and I would stop completely, to reach bus factor >2 (when the 2 top committers have less than 50% of the commits), which at the current commit rate equals in about 5 years. And it would take about 3 years to just push the factor above 1. So even when new people joins the project, they have a really hard time to significantly change the bus factor…

The image above shows the relative share of commits done in the curl project’s git source code repository (as a share of the total amount) by the top 4 commiters from January 1 2010 to July 5 2017 (click for higher resolution). The top dotted line shows the combined share of all four (at 82% right now) and the dark blue line is my share. You can see how my commit share has shrunk from 72% down to 57% over these last 7.5 years. If this trend holds, I’ll have less than 50% of the total commits done in curl in 3-4 years.

At the same time, the thicker light blue line that climbs up into the right is the total number of authors in the git repository, which recently surpassed 500 as you can see. (The line uses the right Y-axes)

We’re approaching 1600 individually named contributors thanked in the project and every release we do (we ship one every 8 weeks) has around 40 contributors, out of which typically around half are newcomers. The long tail is very long and the amount of drive-by just-once contributors is high. Also note how the number 1600 is way higher than the 500 something that has authored commits. Lots of people contribute in other ways.

When we ask our users “why don’t you contribute (more) to the project?” (which we do annually) what do they answer? They say its because 1) everything works, 2) I don’t have time 3) things get fixed fast enough 4) I don’t know the programming language 5) I don’t have the energy.

First as the 6th answer (at 5% 2017) comes “other” where some people actually say they wouldn’t know where to start and so on.

All of this taken together: there are no visible signs of us suffering from having a low bus factor. Lots of signs that people can do things when they want to if I don’t do it. Lots of signs that the code and concepts are understood.

Lots of signs that a low bus factor is not a big problem here. Or perhaps rather that the bus factor isn’t really as low as any tool would calculate it.

What if I…

Do I know who would pick up the project and move on if I die today? No. We’re a 100% volunteer-driven project. We create one of the world’s most widely used software components (easily more than three billion instances and counting) but we don’t know who’ll be around tomorrow to work on it. I can’t know because that’s not how the project works.

Given the extremely wide use of our stuff, given the huge contributor base, given the vast amounts of documentation and tests I think it’ll work out.

Just because you have a large bus factor doesn’t necessarily make the project a better place to ask questions. We’ve seen projects in the past where N persons involved are all from the same company and when that company removes its support for that project those people all go away. High bus factor, no people to ask.

Finally, let me just add that I would of course love to have many more committers and contributors in the curl project, and I think we would be an even better project if we did. But that’s a separate issue.

Mozilla, Work

Denied entry

25 Comments

 – Sorry, you’re not allowed entry to the US on your ESTA.

The lady who delivered this message to me this early Monday morning, worked behind the check-in counter at the Arlanda airport. I was there, trying to check-in to my two-leg trip to San Francisco to the Mozilla “all hands” meeting of the summer of 2017. My chance for a while ahead to meet up with colleagues from all around the world.

This short message prevented me from embarking on one journey, but instead took me on another.

Returning home

I was in a bit of a shock by this treatment really. I mean, I wasn’t treated particularly bad or anything but just the fact that they downright refused to take me on for unspecified reasons wasn’t easy to swallow. I sat down for a few moments trying to gather my thoughts on what to do next. I then sent a few tweets out expressing my deep disappointment for what happened, emailed my manager and some others at Mozilla about what happened and that I can’t come to the meeting and then finally walked out the door again and traveled back home.

This tweet sums up what I felt at the time:

Going back home. To cry. To curse. To write code from home instead. Fricking miserable morning. No #sfallhands for me.

— Daniel Stenberg (@bagder) 26 juni 2017

Then the flood

That Monday passed with some casual conversations with people of what I had experienced, and then…

Someone posted to hacker news about me. That post quickly rose to the top position and it began. My twitter feed suddenly got all crazy with people following me and retweeting my rejection tweets from yesterday. Several well-followed people retweeted me and that caused even more new followers and replies.

By the end of the Tuesday, I had about 2000 new followers and twitter notifications that literally were flying by at a high speed.

I was contacted by writers and reporters. The German Linux Magazine was first out to post about me, and then golem.de did the same. I talked to Kate Conger on Gizmodo who wrote Mozilla Employee Denied Entry to the United States. The Register wrote about me. I was for a moment considered for a TV interview, but I think they realized that we had too little facts to actually know why I was denied so maybe it wasn’t really that TV newsworthy.

These articles of course helped boosting my twitter traffic even more.

In the flood of responses, the vast majority were positive and supportive of me. Lots of people highlighted the role of curl and acknowledged that my role in that project has been beneficial for quite a number of internet related software in the world. A whole bunch of the responses offered to help me in various ways. The one most highlighted is probably this one from Microsoft’s Chief Legal Officer Brad Smith:

I’d be happy to have one of our U.S. immigration legal folks talk with you to see if there’s anything we can do to help. Let me know.

— Brad Smith (@BradSmi) 27 juni 2017

I also received a bunch of emails. Some of them from people who offered help – and I must say I’m deeply humbled and grateful by the amount of friends I apparently have and the reach this got.

Some of the emails also echoed the spirit of some of the twitter replies I got: quite a few Americans feel guilty, ashamed or otherwise apologize for what happened to me. However, I personally do not at all think of this setback as something that my American friends are behind. And I have many.

Mozilla legal

Tuesday evening I had a phone call with our (Mozilla’s) legal chief about my situation and I helped to clarify exactly what I had done, what I’ve been told and what had happened. There’s a team working now to help me sort out what happened and why, and what I and we can do about it so that I don’t get to experience this again the next time I want to travel to the US. People are involved both on the US as well as on the Swedish side of things.

Personally I don’t have any plans to travel to the US in the near future so there’s no immediate rush. I had already given up attending this Mozilla all-hands.

Repercussions

Mark Nottingham sent an email on the QUIC working group’s mailing list, and here follows two selected sections from it:

You may have seen reports that someone who participates in this work was recently refused entry to the US*, for unspecified reasons.

…

We won’t hold any further interim meetings in the US, until there’s a change in this situation. This means that we’ll either need to find suitable hosts in Canada or Mexico, or our meeting rotation will need to change to be exclusively Europe and Asia.

I trust I don’t actually need to point out that I am that “someone” and again I’m impressed and humbled by the support and actions in my community.

Now what?

I’m now (end of Wednesday, 60 hours since the check-in counter) at 3000 more twitter followers than what I started out with this Monday morning. This turned out to be a totally crazy week and it has severally impacted my productivity. I need to get back to write code, I’m getting behind!

I hope we’ll get some answers soon as to why I was denied and what I can do to fix this for the future. When I get that, I will share all the info I can with you all.

So, back to work!

Thanks again

Before I forget: thank you all. Again. With all my heart. The amount of love I’ve received these last two days is amazing.

Update years later

I was eventually given a visa and can travel to the US again.

c-ares

c-ares 1.13.0

The c-ares project may not be very fancy or make a lot of noise, but it steadily moves forward and boasts an amazing 95% code coverage in the automated tests.

Today we release c-ares 1.13.0.

This time there’s basically three notable things to take home from this, apart from the 20-something bug-fixes.

CVE-2017-1000381

Due to an oversight there was an API function that we didn’t fuzz and yes, it was found out to have a security flaw. If you ask a server for a NAPTR DNS field and that response comes back crafted carefully, it could cause c-ares to access memory out of bounds.

All details for CVE-2017-1000381 on the c-ares site.

(Side-note: this is the first CVE I’ve received with a 7(!)-digit number to the right of the year.)

cmake

Now c-ares can optionally be built using cmake, in addition to the existing autotools setup.

Virtual socket IO

If you have a special setup or custom needs, c-ares now allows you to fully replace all the socket IO functions with your own custom set with ares_set_socket_functions.

cURL and libcurl, Network, Web

“OPTIONS *” with curl

7 Comments

(Note: this blog post as been updated as the command line option changed after first publication, based on comments to this very post!)

curl is arguably a “Swiss army knife” of HTTP fiddling. It is one of the available tools in the toolbox with a large set of available switches and options to allow us to tweak and modify our HTTP requests to really test, debug and torture our HTTP servers and services.

That’s the way we like it.

In curl 7.55.0 it will take yet another step into this territory when we finally introduce a way for users to send “OPTION *” and similar requests to servers. It has been requested occasionally by users over the years but now the waiting is over. (brought by this commit)

“OPTIONS *” is special and peculiar just because it is one of the few specified requests you can do to a HTTP server where the path part doesn’t start with a slash. Thus you cannot really end up with this based on a URL and as you know curl is pretty much all about URLs.

The OPTIONS method was introduced in HTTP 1.1 already back in RFC 2068, published in January 1997 (even before curl was born) and with curl you’ve always been able to send an OPTIONS request with the -X option, you just were never able to send that single asterisk instead of a path.

In curl 7.55.0 and later versions, you can remove the initial slash from the path part that ends up in the request by using –request-target. So to send an OPTION * to example.com for http and https URLs, you could do it like:

$ curl --request-target "*" -X OPTIONS http://example.com
$ curl --request-target "*" -X OPTIONS https://example.com/

In classical curl-style this also opens up the opportunity for you to issue completely illegal or otherwise nonsensical paths to your server to see what it does on them, to send totally weird options to OPTIONS and similar games:

$ curl --request-target "*never*" -X OPTIONS http://example.com

$ curl --request-target "allpasswords" http://example.com

Enjoy!

cURL and libcurl

curl doesn’t spew binary anymore

23 Comments

One of the least favorite habits of curl during all these years, I’ve been told, is when users forget to instruct the command line tool where to store the downloaded file and as a direct consequence, curl instead sends a lot of binary “gunk” to the terminal. The end result of that is at best just a busload of weird-looking characters on the screen, but with just a little bit of bad luck it can also lock up the terminal completely or change it in other ways.

Starting in curl 7.55.0 (from this commit), curl will inspect the beginning of each download that has been told to get sent to the terminal (tty!) and attempt to detect and prevent raw binary output to get sent there. The code is only simply looking for a binary zero in the data.

$ curl https://example.com/image.jpg
Warning: Binary output can mess up your terminal. Use "--output -" to tell curl to output it to your terminal anyway, or consider "--output <FILE>" to save to a file.

As the warning message says, there’s an option to use to switch off this emergency check for when you truly know what you’re doing and you don’t need curl to prevent you from doing this. Then you just tell curl explicitly that you want the output to stdout, with “–output -” (or “-o -” for a shorter version):

$ curl -o - https://example.com/binblob.img

We’re eager to get your input and feedback on how this works. We are aware of the risk of false positives for UTF-16 and UTF-32 outputs, but we think they are rare enough to not make this a huge problem.

This feature should be able to drastically reduce the risk for this:

Pipes

(Update, added after the initial posting.)

So many have remarked or otherwise asked how this affects when stdout is piped into something else. It doesn’t affect that! The whole point of this check is to only show the warning message if the binary output is sent to the terminal. If you instead pipe the output to another program or if you redirect the output with >, that will not trigger this warning but will instead continue just like before. Just like you’d expect it to.

cURL and libcurl

curl: read headers from file

Starting in curl 7.55.0 (since this commit), you can tell curl to read custom headers from a file. A feature that has been asked for numerous times in the past, and the answer has always been to write a shell script to do it. Like this:

#!/bin/sh
while read line; do
  args="$args -H '$line'";
done
curl $args $URL

That’s now a response of the past (or for users stuck on old curl versions). We can now instead tell curl to read headers itself from a file using the curl standard @filename way:

$ curl -H @headers https://example.com

… and this also works if you want to just send custom headers to the proxy you do CONNECT to:

$ curl --proxy-headers @headers --proxy proxy:8080 https://example.com/

(this is a pure curl tool change that doesn’t affect libcurl, the library)

cURL and libcurl, Network

curling over HTTP proxy

Starting in curl 7.55.0 (this commit), curl will no longer try to ask HTTP proxies to perform non-HTTP transfers with GET, except for FTP. For all other protocols, curl now assumes you want to tunnel through the HTTP proxy when you use such a proxy and protocol combination.

Protocols and proxies

curl supports 23 different protocols right now, if we count the S-versions (the TLS based alternatives) as separate protocols.

curl also currently supports seven different proxy types that can be set independently of the protocol.

One type of proxy that curl supports is a so called “HTTP proxy”. The official HTTP standard includes a defined way how to speak to such a proxy and ask it to perform the request on the behalf of the client. curl supports using that over either HTTP/1.1 or HTTP/1.0, where you’d typically only use the latter version if you the first really doesn’t work with your ancient proxy.

HTTP proxy

All that is fine and good. But HTTP proxies were really only defined to handle HTTP, and to some extent HTTPS. When doing plain HTTP transfers over a proxy, the client will send its request to the proxy like this:

GET http://curl.haxx.se/ HTTP/1.1
Host: curl.haxx.se
Accept: */*
User-Agent: curl/7.55.0

… but for HTTPS, which should provide end to end encryption, a client needs to ask the proxy to instead tunnel through the proxy so that it can do TLS all the way, without any middle man, to the server:

CONNECT curl.haxx.se:443 HTTP/1.1
Host: curl.haxx.se:443
User-Agent: curl/7.55.0

When successful, the proxy responds with a “200” which means that the proxy has established a TCP connection to the remote server the client asked it to connect to, and the client can then proceed and do the TLS handshake with that server. When the TLS handshake is completed, a regular GET request is then sent over that established and secure TLS “tunnel” to the server. A GET request that then looks like one that is sent without proxy:

GET / HTTP/1.1
Host: curl.haxx.se
User-Agent: curl/7.55.0
Accept: */*

FTP over HTTP proxy

Things get more complicated when trying to perform transfers over the HTTP proxy using schemes that aren’t HTTP. As already described above, HTTP proxies are basically designed only for doing HTTP over them, but as they have this concept of tunneling through to the remote server it doesn’t have to be limited to just HTTP.

Also, historically, for decades people have deployed HTTP proxies that recognize FTP URLs, and transparently handle them for the client so the client can almost believe it is HTTP while the proxy has to speak FTP to the remote server in the other end and convert it back to HTTP to the client. On such proxies (Squid and Apache both support this mode for example), this sort of request is possible:

GET ftp://ftp.funet.fi/ HTTP/1.1
Host: ftp.funet.fi
User-Agent: curl/7.55.0
Accept: */*

curl knows this and if you ask curl for FTP over an HTTP proxy, it will assume you have one of these proxies. It should be noted that this method of course limits what you can do FTP-wise and for example FTP upload is usually not working and if you ask curl to do FTP upload over and HTTP proxy it will do that with a HTTP PUT.

HTTP proxy tunnel

curl features an option (–proxytunnel) that lets the user forcible tell the client to not assume that the proxy speaks this protocol and instead use the CONNECT method with establishing a tunnel through the proxy to the remote server.

It should of course be noted that very few deployed HTTP proxies in the wild allow clients to CONNECT to whatever port they like. HTTP proxies tend to only allow connecting to port 443 as that is the official HTTPS port, and if you ask for another port it will respond back with a 4xx response code refusing to comply.

Not HTTP not FTP over HTTP proxy

So HTTP, HTTPS and FTP are sent over the HTTP proxy fine. That leaves us with nineteen more protocols. What happens with them when you ask curl to perform them over a HTTP proxy?

Now we have finally reached the change that has just been merged in curl and changes what curl does.

Before 7.55.0

curl would send all protocols as a regular GET to the proxy if asked to use a HTTP proxy without seeing the explicit proxy-tunnel option. This came from how FTP was done and grew from there without many people questioning it. Of course it wouldn’t ever work, but also very few people would actually attempt it because of that.

From 7.55.0

All protocols that aren’t HTTP, HTTPS or FTP will enable the tunnel-through mode automatically when a HTTP proxy is used. No more sending funny GET requests to proxies when they won’t work anyway. Also, it will prevent users from accidentally leak credentials to proxies that were intended for the server, which previously could happen if you omitted the tunnel option with a few authentication setups.

HTTP/2 proxy

Sorry, curl doesn’t support that yet. Patches welcome!

curl, open source and networking