November 1st

Japanese kana keyboard

Pinyin has become strange to me since I have memorized double spelling, which also affects the Romanization of Japanese. Moreover, it's not pleasant to use romanization when you need to input a large amount of Japanese (although I don't need it in real life). Therefore, I wanted to learn the kana input method a few months ago, but when I saw the disordered keys, I had to quit. Today, I finally tried it, but found that it was surprisingly easy to remember the key position. It took me less than two hours to remember it.

Although the distribution of pseudonyms seems irregular, it's a bit catchy to read, even with some meanings. Like reciting five character quatrains, I memorize the order of pseudonyms, and then correspond with the keys one by one. Remove the function key and the ~ key used to toggle the full width, starting with the number 1 and ending/ending

Because the QWERTY keyboard that we usually use is different from JIS keyboard, there is no position of and , which need to be typed with shift + 'and shift +] respectively. At the same time, the key position of quotation marks also changes. The upper and lower quotation marks are Shift + = and shift + [. Although you can find the key position without looking at the figure, it takes a long time, and you need to practice more if you want to use it freely.

October 31st

Don't despise anyone

The MLB of 2019 season finally came to an end on the last day of October, winning all four away games for the first time in the world series, and 36 year old Kendrick also set a record, becoming the first player to hit the whole base in both playoffs. In NLDS's fifth game against Dodge, Kendrick knocked out a grand slam in the first half of the 10th inning; in today's WS seventh game against astronauts, he knocked out two more shots in the first half of the 7th inning. These two home runs not only opened the door to victory for the team, but also created the history of the team.

Kendrick is not a star player, with an annual salary of more than 3 million. He made many mistakes against Dodge, and his position as a second baseman after entering the world series is no longer stable. But it is such a player who becomes the one who breaks the balance and the one who gives the opponent a fatal blow. When looking back on the national championship, people remember not his mistakes, but the arc of flying off the field after swinging the stick.

October 29th

Love instructions

A song I like to listen to recently is " . I should have heard this song before. It seems that it's not the name, but I seldom read the song's name when I listen to it, and I may remember it wrong. This is a sweet love song, and I don't want to fall in love, just feel happy after listening. Although I'm tired of listening to many joyful songs, sometimes it's good to listen to some, just like the love songs of Liang Jingru. When you are anxious, no matter what music you listen to, it's lucky to have a happy song.

Memory of little lily

In the past, I wrote some articles on little lily at first, but also some comments, though not many. Because xiaobaihe hasn't opened an out of school visit yet, thinking that she can't log in any more one day, she decides to transfer the comments of her friends, which is also a little commemoration.

There is no good way to migrate the comments of Lili. You can only register them one by one. If you fill in the comments directly in the background of leancloud, you can't change the creation time of the comments, but you can customize the createdat field in the external import file. Leancloud supports JSON and CSV formats. I chose a relatively simple CSV. The first line is the field type, the second line is the field name, and the content starts from the third line. I have selected several necessary fields, first input them with Google sheets, and then download them as CSV files, which is more convenient.

One of the more annoying things is that the time format of leancloud is ISO 8601 format of UTC time zone, while that of little lily is May 30 11:10:21 2014. Although the number of comments is not large and manual modification will not take too long, the repetitive and monotonous work is annoying after all, so I wanted to do it in one step with R and wrote a simple conversion function, but in fact, it did not improve the efficiency.

October 27th

Batch processing pictures with R

For some reason, these two days, I changed the valine comment to the version of desert, which makes it easier to change the expression pack. I downloaded a set of Rick's stickers on the telegram, but the original picture is a little large, and the file name needs to be changed. I don't know what batch processing tools are available, and the requirements are simple, so I chose to use r to complete the work. However, my modified file name can't reflect the content of facial expression, and I really don't know how to name some facial expressions. In addition, the reduced image is not very clear, so I finally gave up replacing the facial expression pack.

Burning GCP

I accidentally opened Google cloud and saw that I had spent more than 3200 yen this month. I was really shocked. When I first tried GCP, I opened an instance on the Japanese server to run rstudio server, but it was almost unused, so there was no cost. Later, I opened another SQL instance and wordpress blog, and the money burned was probably from these two services.

Open the bill, but I can't understand the service represented by SKU. The most expensive product is cloud SQL running on the Asia Pacific server instance. I thought it was a WP database at first, but now I think it should be a storage bucket for SQL. The second is the instance running rstudio server. The cost seems to be higher than