Now github orders files the wrong way. I suggest solving this by adding a leading zero to numbers < 10

Before:

part1.md
part10.md
part11.md
part12.md
part13.md
part14.md
part2.md
part3.md
part4.md
part5.md
part6.md
part7.md
part8.md
part9.md

After

part01.md
part02.md
part03.md
part04.md
part05.md
part06.md
part07.md
part08.md
part09.md
part10.md
part11.md
part12.md
part13.md
part14.md

original

2.2 A world of options
HTTP 1.1's nature of having lots of tiny details and options available for later extensions
have grown a software ecosystem where almost no implementations ever implement
everything – and it isn't even really possible to exactly tell what “everything” is. This has
lead to a situation where features that were initially little used saw very few
implementations and those who did implement the features then saw very little use of
them. It caused an interoperability problem when then much later on clients and servers
have started to increase the use of some such features. HTTP Pipelining being a primary example of such a one.

Then much later on, it caused an interoperability problem when clients and servers started to increase the use of such features. HTTP Pipelining is a primary example of such a feature.

Add descriptions on how to activate and use HTTP/2 with these popular servers, based on how it is done in the existing (patches/pre-release) versions.

I suggest add mention a SPDY Indicator plugin for chrome for check if server has support at http/2

https://chrome.google.com/webstore/detail/spdy-indicator/mpbpobfflnpcgagjijhmgnchggcjblin

Has a similar plugin on Firefox but at the moment no has support at http/2

https://addons.mozilla.org/pt-BR/firefox/addon/spdy-indicator/

Over the years web developers and web development environments have gather a full toolbox of tricks and tools to work around problems with HTTP 1.1, some of them as I outlined in the beginning of this document.

gathered

A problem here is of course that web sites and web developers need to develop and deploy for a world that in the short term at least, will have both HTTP1.1 and http2 clients as users and to get maximum performance for all users can be challenging without having to offering two different front-ends.

offer

For these reasons alone, I suspect there will be some time before we will see the full potentials of http2 being used.

potential of http2 being reached.

At http://http2-explained.haxx.se/content/ there is no an option for Portuguese. Is it right?

Thanks,

Regarding the news about "The secret message hidden in every HTTP/2 connection" it could be fun to have a section/chapter with trivialities like this. mnot's comment on this particular thing is here: http://blog.jgc.org/2015/11/the-secret-message-hidden-in-every.html#5252055626418281054

I started the Swedish translation, mostly to get the ball rolling. So far it is only part1 with a non-translated document history (I find that part more or less unnecessary to translate anyway since those versions never existed in Swedish).

There are ten different frames
define in the http2 spec and the
two perhaps most fundamental
ones that map HTTP 1.1 features
are DATA and HEADERS.

define -> defined

Convert the ODF version into a HTML version that can be hosted on a single web page and in one page per chapter.

The work has began. I store the images in a separate directory (https://github.com/bagder/http2-explained/tree/master/images ) and I name the HTML directory after the language of the text (https://github.com/bagder/http2-explained/tree/master/en ) to allow more (future) versions in case translators want to use HTML too.

I have not considered CSS or other styling issues and I suck at that anyway.

... It should then ideally also include details on how to deal with TLS decryption

flat odf is supposedly better for version control so let's try to use this for upcoming versions:
https://wiki.documentfoundation.org/Libreoffice_and_subversion

Problem

Mentioning colours in text referring to diagrams displayed on grayscale ebook readers makes it hard to understand the diagram since the colours are indistinguishable.

Solution

Don't use colours to differentiate but use symbols or changes in line drawing. For instance a dotted line vs a connected line. This is easy to understand on a gray scale ereader (in my case an older Kindle).

Details

In '2.4 Transfer sizes and number of objects' you're referring to specific colors:

"It shows the growth of the total transfer size (in green) and the total number of requests used on average (in red) to serve the most popular web sites in the world, and how they have changed over the last four years."

ps: Thanks for sharing your knowledge!

When I do HTTP/2 presentations I get a lot of questions about server push so I think it is worthwhile to expand that section and describe:

1 - the client can reject them at once if it wants to (and it has a small initial window)
2 - the contents will not end up in the browser cache immediately unless the resource is otherwise used
3 - no, a server cannot DOS a client with this in anyway more than it already could using other means

The subject about mandatory TLS has caused much hand- waiving and agitated voices in mailing lists and meetings – is it good or is it evil?

waiving -> waving

The group is named HTTPbis where the “bis” part comes from the Latin adverb for "two".

I don't know much Latin, but "two" is an adjective in every language I know.

Hi Daniel,
Not sure if Gitbook is limited regarding the search feature.
I tried Chrome 45 and Safari on Mac, it doesn't highlight the term you search for, though it will limit the scope to the pages that have the searched for term.

-Olivier.

Everyone who attends and says something have the same weight and
chance to affect the outcome and everyone is counted as humans and individuals, little
care is given to which companies the individuals work for.

have -> has

Everyone who attends and says something has the same weight and
chance to affect the outcome and everyone is counted as humans and individuals, little
care is given to which companies the individuals work for.

Hi Daniel,
I see the Gitbook published book doesn't have any page breaks when converted to PDF. Not sure if it's a Gitbook limitation or a formatting issue.
Some people may like to download and print the PDF.

SPDY3
is a protocol that was developed and spearheaded by Google. They certainly
developed it in the open and invited everyone to participate and but it was obvious that
they had huge benefits by being in control over both a popular browser implementation
and a significant server population with well-used services.

"and but" -> "but"

SPDY3
is a protocol that was developed and spearheaded by Google. They certainly
developed it in the open and invited everyone to participate but it was obvious that
they had huge benefits by being in control over both a popular browser implementation
and a significant server population with well-used services.

When the HTTPbis group decided it was time to start working on http2. SPDY had already
proven that it was a working concept.

. -> ,

When the HTTPbis group decided it was time to start working on http2, SPDY had already
proven that it was a working concept.

The link for the french translation at http://daniel.haxx.se/http2/ is now outdated as it points to the old fodt generated file.

6.5. Сжатие заголовков

Paragraph
Как я уже упоминал, в то время как число объектов на странице увеличивается, использование cookie и размер запросов также продолжают расти. Cookie также должны быть включены во все запросы, практически всегда одинаковые на протяжение множества запросов.

Need replace "на протяжение множества запросов" to "на протяжении множества запросов"

Since it may have an impact and reduce connections even more than just the single TCP connection per host.

Hi,
First of all, I would like to thanks, this book is very cool and helped me a lot.

What do you think about to translate it for any languages?

Thanks!

from github description:
"http2 explained is a detailed document explainining and documenting the new protocol"

should be explaining (unless it was on purpose)

12.1. QUIC

Протокол QUIC (Quick UDP Internet Connections, Быстрые UDP Интернет Соединений) Гугла – это интересный эксперимент, проведённый в том же стиле, как и с SPDY. QUIC – это замена TCP + TLS + HTTP/2, реализованная поверх UDP.

"Быстрые UDP Интернет Соединений" replace to "Быстрые UDP интернет соединения"

Chapter 5.4 title isn't translated:
5.4. http2 for http://
Could be "5.4. http2 et http://"

@machour if you have time to look at it

Minor typo in the text of section 11.4 in the guide:
"To tell curl to use http2, either plain text or over TLS, you use the –http2 option (that is
“dash dash http2”). curl still defaults to HTTP/1.1 so the extra option is necessary when
you want http2."

The "-http2 option" bit contains only one hyphen/dash and should be changed to line up with the text

8.4.3. “The protocol is only useful big sites”
version 1.10

Looks like a "for" is missing. s/useful big/useful for big/ ?

If I guess correctly that "voilá" on English version pdf page 9 is the French word, then it should be spelled "voilà". There is no a-acute letter in French.

original

HTTP 1.1 has a hard time to really take full advantage of all the powers and performance
that TCP offers

new

HTTP 1.1 has a hard time really taking full advantage of all the powers and performance that TCP offers

Hey @jabiinfante, I noticed your awesome merge in the is the 'es' subdir! Is the Spanish markdown version now "good enough" to be the one to link to from the web page at http://daniel.haxx.se/http2 ?

So what's http2 set out to do then? Where's the boundaries for what the HTTPbis group
was set out to do?

Where's -> Where are
was set -> set

So what's http2 set out to do then? Where are the boundaries for what the HTTPbis group set out to do?

There's no more minor versions in http2.

There are no more minor versions in http2.

HTTP initially said that a client was allowed to use maximum two TCP connections for each host. So, in order to not violate the spec clever sites simply invented new host names and voilá

Should be "to use at most two TCP connections", and the accent is "voilà".

Also it would be nice to be more specific, e.g. "The HTTP 1.1 specification section 8.1.4 said 'A single-user client SHOULD maintain AT MOST 2 connections with any server or proxy.' " I don't see any mention of the two connection limit in RFC 1945 for HTTP 1.0.

http2 reduces the number of necessary network round-trips and it avoid the head of line blocking dilemma completely with multiplexing and fast discarding of unwanted streams.

avoid -> avoids

The final trick I'll mention that sites owners to do perform better in browsers is often
referred to as “sharding”

change "to do" to "do to"

Since tests have proven that stream dependencies can have a significant impact on performance or time to first-draw, I think it is worth expanding the description of that.

The annoyance to developers and the hoops everyone has to go through to do this is of course “just” pain for the humans involved and not seen in any performance numbers...

I'm not a native speaker but this sentence is difficult to understand. What is not seen in any performance numbers? The annoyance, I guess (after reading this sentence 10 times)? On my first read I understood that this trick did not improve performance.

Thanks for the book!

I suggest adding a small link to the next file in the bottom for easier reading on github. It's obviously better to use relative links to next chapter, e.g.

[next part →](./part02.md)

producing
next part →

Not sure how you would feel about that, but I'm using a slow network connection and having to download the now outdated English and French fodt files seems kind of pointless now.

Maybe the presentation fodp file could be vacuum flushed too ? Everything would anyway reside for ever in the git history.

Spanish should be kept until migration is done though.

I think ALPN means Application-Layer Protocol Negotiation instead of Application Level Protocol Negotation. That is at least what RFC 7301 says.

By the way, thanks for that excellent summary.

The cover page link takes the user to "http.//daniel.haxx.se/http2/", but should be "http://daniel.haxx.se/http2/" (I discovered this error when I clicked on the link in a browser).

I believe my switching on the multi-language support or something broke the images. They are now rendered with wrong path and thus aren't visible.

I put the images in a common directory so that translations can use the same pics but apparently I need to do something different to make it work. I contacted gitbook and asked about it.

For every stream both ends has to tell the peer that it has more room to fit incoming data in, and the other end is only allowed to send that much data until the window is extended.

For every stream, both ends have

In 6.5 Header Compression, paragraph 2:

A series of almost identical something begs for compression.

Should be:

A series of almost identical somethings begs for compression.

See http://help.gitbook.com/format/cover.html

"A cover is specified by a cover.jpg file, a cover_small.jpg can also exist as a smaller version of the cover. The cover should be a JPEG file."

The specific sizes to use are: 1800x2360 and 200x262

8.3 says:
"curl and libcurl has its implementation based on the separate http2 library called nghttp2 and supports plain-text http2..."

plain-text http2?! http2 is binary and it cannot be plain-text, this sentence is confusing. You mean non-TLS http2, right?

The HTTPbis working group was formed during the summer of 2007 and was set out to
do an updated of the HTTP 1.1 specification

should be "update" instead of "updated"

Sure you can be careful about
picking line so that you pick the
one you really believe is the
correct one, and at times you
can even start a new line of
your own but in the end you
can't avoid making a decision
and once it is made you cannot
switch lines.

"line picking" sounds more natural to me than "picking line"

On page 6, the line graph (presumably showing objects/page and load size) includes no key to specifically state that it is the case.

Just for best practice and all.