-
```
The word "嘅" (U+5605) is composed by a phonetic component "既" with a
radical "口" (the mouth), i.e. 嘅 = 口 + 既.
Similar words are 溉, 概 and 慨, where 溉 = 氵+ 既 , 概 = 木 + 既, and
慨 = 忄 + 既.
There are…
-
Hi there,
I'm from Taiwan, where we use Traditional Chinese. The CotEditor provides some non-Unicode encodings for Trad. Chinese, but the most common ones, Big5 series, is missing.
Here is what Text…
-
Hi! It looks like is using a license "HKSCS" that doesn't have a text file in either this overlay or the main tree. Could it be that it should be HKSCS-2001, instead?
-
```
The font face used for 0.4.4 is not working on N95-4. Firmware is newest
available for model (v20.2.005). Text appears as empty boxes with the
occasional pixelfont-style capital letters.
Screen…
-
```
The font face used for 0.4.4 is not working on N95-4. Firmware is newest
available for model (v20.2.005). Text appears as empty boxes with the
occasional pixelfont-style capital letters.
Screen…
-
http://www.edu.tw/FILES/SITE_CONTENT/M0001/BIAU/f04.htm?open
According to Taiwan Standard, 又 should be closed.
When something in 又 (aka. 叉、蚤), 又 should be opened.
Glyph '又' of Source Han Sans looks …
-
When using the following in emacs:
```
#
GET http://localhost:8083/test/1
Accept: text/xml
Accept-Charset: utf-8
```
emacs actually sends
```
GET /test/1 HTTP/1.1
MIME-Version: 1.0
Connection: keep…
-
When I try to set locale, I've got a strange result. State says locale added successfully, although logs says locale-gen failed with exit code 1.
```
ar_AE.UTF-8 UTF-8:
locale.present
```
And here…
holms updated
9 years ago
-
Sometimes people needs to write Japanese name in Traditional Chinese.
The Japanese Jinmeiyō kanji can be divided into 4 catalogues.
1. The character is already included in Big5 and HKSCS
Do nothin…
-
https://github.com/nathancolgate/spreadsheet/blame/master/lib/spreadsheet/excel/internals.rb
These are encodings on MRI 2.1.2
``` ruby
Encoding.list
=> [
#,
#,
#,
#,
#,
#,
#,
#,
#,
#,
#,
…