Thadeu Lima de Souza Cascardo [off-list ref] writes:
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
Thanks.
+Você também pode dar ao 'git-log' um "intervalo" de commits onde o
+primeiro não é necessariamente um ancestral do segundo; por exemplo, se
+as pontas dos ramos "stable" e "master" divergiram de um commit
+comum algum tempo atrás, então
+
+-------------------------------------
+$ git log stable..experimental
+-------------------------------------
+
+irá listas os commits feitos no ramo experimental mas não no ramo
+stable, enquanto
+
+-------------------------------------
+$ git log experimental..stable
+-------------------------------------
+
+irá listar a lista de commits feitos no ramo stable mas não no ramo
+experimental.
+
I think you would want to update this part to match what you did in your
[PATCH 1/2 v2].
By the way, I think your MUA sent quoted-printable UTF-8 but somewhere
between your keyboard and vger the message was marked with content-type
charset=ISO-8859-1); I fixed it up when quoting the above.
I am somewhat worried about the way how this translation will be
maintained to keep in sync with the authoritative English version.
Narita-san (CC'ed) who translated the document to Japanese did this:
gittutorial(7)
==============
// = gittutorial(7)
NAME
----
// == NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or newer)
// gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
and the idea seems that without // (comments in AsciiDoc markup) it
matches the English copy, and after passing sed -ne 's|^// ||p' it yields
Japanese version. Narita-san's translation can be seen at
http://github.com/yasuaki/git-manual-jp.git/Documentation
if anybody is interested.
With this format, merging upstream changes may not work as smoothly as it
could be, but at least you can check which part of your translation is
based on a stale copy with something like this arrangement.
I am wondering if it would be a good idea to extend Narita-san's scheme so
that we can keep a single source, perhaps like:
= gittutorial(7)
// ja = gittutorial(7)
// pt = gittutorial(7)
== NAME
// ja = NAME
// pt = NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or ...
// ja gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
// pt gittutorial - Um tutorial de introdução ao git (para versão 1....
Then whenever somebody makes a change to the English version, he can and
should also mark the corresponding translated versions "stale", so that it
is easier to spot by translators.
diff --git a/gittutorial.txt b/gittutorial.txt
index 4478300..02d67d3 100644
--- a/gittutorial.txt
+++ b/gittutorial.txt
@@ -4,7 +4,6 @@
== NAME
// ja = NAME
// pt = NAME
-gittutorial - A tutorial introduction to git (for version 1.5.1 or n...
-// ja gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
-// pt gittutorial - Um tutorial de introdução ao git (para versão 1....
-
+gittutorial - A tutorial introduction to git (for version 1.6.3 or n...
+// **stale** ja gittutorial - git チュートリアル (バージョン 1.5.1 ...
+// **stale** pt gittutorial - Um tutorial de introdução ao git (par...
As long as all the translations use the same encoding (I think UTF-8 is
the only practical choice for this), keeping translated strings in a
single file would be doable.
I however am not sure how practical it would be to force people to look at
the *.txt version of document, only 1/n lines of which is now readable by
him (if you are like a typical American who understands only English ;-).
Thoughts?
On Mon, Jun 29, 2009 at 09:08:00AM -0700, Junio C Hamano wrote:
Thadeu Lima de Souza Cascardo [off-list ref] writes:
quoted
Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>
Thanks.
quoted
+Você também pode dar ao 'git-log' um "intervalo" de commits onde o
+primeiro não é necessariamente um ancestral do segundo; por exemplo, se
+as pontas dos ramos "stable" e "master" divergiram de um commit
+comum algum tempo atrás, então
+
+-------------------------------------
+$ git log stable..experimental
+-------------------------------------
+
+irá listas os commits feitos no ramo experimental mas não no ramo
+stable, enquanto
+
+-------------------------------------
+$ git log experimental..stable
+-------------------------------------
+
+irá listar a lista de commits feitos no ramo stable mas não no ramo
+experimental.
+
I think you would want to update this part to match what you did in your
[PATCH 1/2 v2].
Well remembered. Thanks.
By the way, I think your MUA sent quoted-printable UTF-8 but somewhere
between your keyboard and vger the message was marked with content-type
charset=ISO-8859-1); I fixed it up when quoting the above.
I am going to take a look at it.
I am somewhat worried about the way how this translation will be
maintained to keep in sync with the authoritative English version.
Narita-san (CC'ed) who translated the document to Japanese did this:
gittutorial(7)
==============
// = gittutorial(7)
NAME
----
// == NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or newer)
// gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
and the idea seems that without // (comments in AsciiDoc markup) it
matches the English copy, and after passing sed -ne 's|^// ||p' it yields
Japanese version. Narita-san's translation can be seen at
http://github.com/yasuaki/git-manual-jp.git/Documentation
if anybody is interested.
With this format, merging upstream changes may not work as smoothly as it
could be, but at least you can check which part of your translation is
based on a stale copy with something like this arrangement.
I am wondering if it would be a good idea to extend Narita-san's scheme so
that we can keep a single source, perhaps like:
= gittutorial(7)
// ja = gittutorial(7)
// pt = gittutorial(7)
== NAME
// ja = NAME
// pt = NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or ...
// ja gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
// pt gittutorial - Um tutorial de introdução ao git (para versão 1....
Then whenever somebody makes a change to the English version, he can and
should also mark the corresponding translated versions "stale", so that it
is easier to spot by translators.
diff --git a/gittutorial.txt b/gittutorial.txt
index 4478300..02d67d3 100644
--- a/gittutorial.txt
+++ b/gittutorial.txt
@@ -4,7 +4,6 @@
== NAME
// ja = NAME
// pt = NAME
-gittutorial - A tutorial introduction to git (for version 1.5.1 or n...
-// ja gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
-// pt gittutorial - Um tutorial de introdução ao git (para versão 1....
-
+gittutorial - A tutorial introduction to git (for version 1.6.3 or n...
+// **stale** ja gittutorial - git チュートリアル (バージョン 1.5.1 ...
+// **stale** pt gittutorial - Um tutorial de introdução ao git (par...
As long as all the translations use the same encoding (I think UTF-8 is
the only practical choice for this), keeping translated strings in a
single file would be doable.
I however am not sure how practical it would be to force people to look at
the *.txt version of document, only 1/n lines of which is now readable by
him (if you are like a typical American who understands only English ;-).
Thoughts?
I think that using something like po would be better. There are tools
that can extract and update the template messages from many differente
sources. Adapting them to produce a template file from gittutorial.txt
would allow translators to verify how stale their translations are and
much smoother merges. How about that?
Regards,
Cascardo.
Junio C Hamano [off-list ref] writes:
Thadeu Lima de Souza Cascardo [off-list ref] writes:
quoted
+Você também pode dar ao 'git-log' um "intervalo" de commits onde o
+primeiro não é necessariamente um ancestral do segundo; por exemplo, se
+as pontas dos ramos "stable" e "master" divergiram de um commit
+comum algum tempo atrás, então
+
+-------------------------------------
+$ git log stable..experimental
+-------------------------------------
I am somewhat worried about the way how this translation will be
maintained to keep in sync with the authoritative English version.
Narita-san (CC'ed) who translated the document to Japanese did this:
gittutorial(7)
==============
// = gittutorial(7)
NAME
----
// == NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or newer)
// gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
and the idea seems that without // (comments in AsciiDoc markup) it
matches the English copy, and after passing sed -ne 's|^// ||p' it yields
Japanese version. Narita-san's translation can be seen at
http://github.com/yasuaki/git-manual-jp.git/Documentation
if anybody is interested.
With this format, merging upstream changes may not work as smoothly as it
could be, but at least you can check which part of your translation is
based on a stale copy with something like this arrangement.
I am wondering if it would be a good idea to extend Narita-san's scheme so
that we can keep a single source, perhaps like:
= gittutorial(7)
// ja = gittutorial(7)
// pt = gittutorial(7)
== NAME
// ja = NAME
// pt = NAME
gittutorial - A tutorial introduction to git (for version 1.5.1 or ...
// ja gittutorial - git チュートリアル (バージョン 1.5.1 以降用)
// pt gittutorial - Um tutorial de introdução ao git (para versão 1....
Then whenever somebody makes a change to the English version, he can and
should also mark the corresponding translated versions "stale", so that it
is easier to spot by translators.
[...]
I however am not sure how practical it would be to force people to look at
the *.txt version of document, only 1/n lines of which is now readable by
him (if you are like a typical American who understands only English ;-).
Thoughts?
Somebody here (or on #git channel) pointed that there is po4a[1-4]
project using gettext to translate documentation.
[1] http://po4a.alioth.debian.org/
[2] https://launchpad.net/po4a
[3] http://freshmeat.net/projects/po4a
[4] http://www.ohloh.net/p/po4a
--
Jakub Narebski
Poland
ShadeHawk on #git