10:33 a.m.
Hi,
Attached below is a copy of updated zh_CN translation for PowerTOP
(master).
This transltion, like every other translation, will mess up the layout
of tabs because PowerTOP does not count the displayed width of strings
correctly.
In
https://github.com/fenrus75/powertop/blob/ea5c678/src/display.cpp#L137-L145,
PowerTOP renders a tab and decides the position of the next from the
"width" or the string, but it's neither using the correct string nor
using the correct calculation for width. Instead of taking the length of
the rendered translation, PowerTOP uses the length of the original
English string. Instead of taking a width from
`wcswidth(tab_translations[tab_names[i]].c_str())`, PowerTOP uses
`strlen()`, breaking on characters that take multiple (typically two)
columns on screen like all of CJK and emoji.
We are currently attempting on a fix to this issue, and current progress
may be followed here: https://github.com/AOSC-Dev/powertop/.
Best Regards
Mingcong Bai
1:04 p.m.
New subject: [Powertop] [po] Updated zh_CN translation
On 2018-03-02 18:33, Mingcong Bai wrote:
We are currently attempting on a fix to this issue, and current
progress may be followed here: https://github.com/AOSC-Dev/powertop/.
Hi folks,
I think I have figured out the fix, but I can't be sure about the
patches
until Bai does a test compilation given how messy the strlen() calls
are.
Help yourself with the attachments if you want to see them this early. I
will
do a more "conventional" email patch-thread after Bai tests it.
Those are fairly mechanical tasks, so take it as whatever license you
want.
For functions added/rewritten entirely on my own, consider them WTFPL.
Amend the changes as you like.
By the way, please remember to update your LINGUAS file so make knows to
compile and ship the translations. Bai did that in:
1. https://github.com/AOSC-Dev/powertop/commit/ca4b2d9
2. https://github.com/AOSC-Dev/powertop/commit/d9ccc30
3. https://github.com/AOSC-Dev/powertop/commit/9347162 (necessary evil…)
--
Best Regards,
Mingye Wang
7:14 p.m.
New subject: [Powertop] [po] Updated zh_CN translation
Updated Mingye Wang's patches to fix build - PowerTOP has been built and
tested and columns are now aligned correctly when using zh_CN.
Best Regards,
Mingcong Bai
在 2018年03月02日 20:04, arthur2e5(a)aosc.io 写道:
On 2018-03-02 18:33, Mingcong Bai wrote:
>
> We are currently attempting on a fix to this issue, and current
> progress may be followed here: https://github.com/AOSC-Dev/powertop/.
Hi folks,
I think I have figured out the fix, but I can't be sure about the patches
until Bai does a test compilation given how messy the strlen() calls are.
Help yourself with the attachments if you want to see them this early.
I will
do a more "conventional" email patch-thread after Bai tests it.
Those are fairly mechanical tasks, so take it as whatever license you
want.
For functions added/rewritten entirely on my own, consider them WTFPL.
Amend the changes as you like.
By the way, please remember to update your LINGUAS file so make knows to
compile and ship the translations. Bai did that in:
1. https://github.com/AOSC-Dev/powertop/commit/ca4b2d9
2. https://github.com/AOSC-Dev/powertop/commit/d9ccc30
3. https://github.com/AOSC-Dev/powertop/commit/9347162 (necessary evil…)
--
Best Regards,
Mingye Wang
6:49 a.m.
New subject: [Powertop] [PATCH 0/2] Align multi-column characters properly (SPLIT [po] Updated zh_CN translation)
Hi folks,
On 18/03/03 03:14, Mingcong Bai wrote:
Updated Mingye Wang's patches to fix build - PowerTOP has been
built and
tested and columns are now aligned correctly when using zh_CN.
Best Regards,
Mingcong Bai
I will post the patches below, as I just found something else to revise.
Here are some quick answers to typical code reviewer questions:
Q. What strings does measure_string fail on?
A. Strings with invalid multibyte sequences (-3), non-printable
characters such as newlines and tabs (-1), NULL (-2), and huge strings
that fail malloc (-2).
Q. Why is measure_string returning an int instead of a size_t?
A. I blame wcswidth.
Q. What does measure_string depend on?
A. malloc/free, mbsrtowcs, and wcswidth.
Q. Is align_string slower now?
A. Yes. It needs to do a full encoding conversion in measure_string
instead of just counting.
Q. Are you making a more "naive" version of align_string?
A. Not now. But note that the unit-printing parts (watts, rpm...)
generally only involve printable ASCII, so strlen() will suffice for
them. It should not matter too much.
Q. Is the "ctr" variable useless in cpu.cpp now?
A. Yes, but it still should sync up with the actual column count.
Things we want you all to fix up that aren't included in the patch:
1. LINGUAS, or we will never see the translations used.
2. The use of #include <stdfoo.h> in C++ where #include <cstdfoo> is
available. The C++ variants of these headers work much better (you don't
have to memset the mbstate_t, for example) in C++.
3. Use some C++ strings maybe? It's not that bad for me as compilers
know how to optimize out stuff like "if (strlen(x))", but those new
stuff sure are much easier to use.
--
Regards,
Mingye Wang
1660
days inactive
1661
days old
3 comments
3 participants
participants (3)
-
arthur2e5＠aosc.io -
Mingcong Bai -
Mingye Wang (Artoria2e5)