What is your opinion on the nostarch manga guides?
What is your opinion on the nostarch manga guides?
Other urls found in this thread:
ipfs.io
en.wikipedia.org
twitter.com
It does what it aims to do.
Databases one is surprisingly good.
My ten year old loves them.
Hell yea.
You should give that ten year old back. I'm sure its family misses it.
What's with their 1950s graphical aesthetic (the manga guides maybe being an exception)?
The illustrators like that aesthetic. I'm sure that if they could hire American comic book artists, the books they publish will feature comic style art.
I really enjoy the look of the covers.
I kind of want to get the GNU make book from them.
Where do I pirate them? Checked all deepweb libraries known to me and couldn't find anything.
Piracy happens on the high seas.
it makes me want to hang myself
Any Manga PDFs?
Too slow, too simplified. If you have to stoop that low, might as well choose {Language} No Ehon (literally a picture book).
Otherwise, Codecademy, Two scoopes of Django, Learn You A Haskell and The C programming language is enough.
Agreed, it's a useful intro read.
I'll post them when I get home. I'm surprised no one else here has them.
Mind if I ask, do you have The Elephant Book and the Alien book in Japanese?
I can't read moon so sorry, I don't.
If only all animefaggots were this civilized. Polite sage, only because I haven't seen a moe piece of trash png in this thread yet.
Any news?
Gay. Like OP.
You can find a lot of manga guide pdfs on libgen
Thank you user!
Here is the whole 13 book collection:
ipfs.io
Learning is done best with anime girls.
Ne~
What is the best basic environment to practice PL-SQL, (preferably as lightweight as possible)?
My isp forbids me from connecting to ipfs.
I guess SQLite, it uses a single file for the database, so there is practically nothing to set up. Just create a file, play with it, and throw it away when you are done.
Annoying as usual.
Can SQLite even into PL/SQL?
Anime is all about simplicity, for greatest effect.
A very efficient medium, or lazy. Matter of perspective.
There's a product called StepSqlite that's supposed to compile a subset of PL/SQL for SQLite, but I don't think that's was user was referring to. I don't think he realized that PL/SQL isn't the same as SQL.
en.wikipedia.org
p.sad you can't read a proper book and have to resort to shitty chinese comic books tbh
Buy cheapest VPS and use ssh -D. Takes 15 minutes and 3$.
This one's sounds good but I don't think it's scanned.
What is the actual primary significant content of this image? I'm about as confused as a bot would probably be parsing it and trying to make deeper sense of it.
please don't learn circuit theory from a manga garbage
t. (((textbook publisher)))
lol
they fucked up the formatting pretty bad here
took only 2 minutes of reading to find it, so there's probably a lot more
you mean the ipfs.io website?
just use tor and be done
I mean, it's very unlikely he will actually read the manga let alone apply that knowledge. Not like he's gonna earn money either. Nothing wrong with learning from that book imho.
Yeah, I just read SQL, not the PL part before that. Sage for being an idiot.
What software is recommended to use to practice basic database stuff? It seems like a vast and complex area of technology with many different database management systems and many different SQL variants, so it's hard to figure out what to even start with. What free software is suitable for intro level practice (preferably one that's universal enough and not tied too much to a specific DBMS/SQL implementation, and one that does not bog you down with too much details and complexity all at once).
idk how this is even a problem
just dive right in into the most popular ones right now to get a feel of things.
Such as? What in particular would you suggest?
Whats wrong with that? Its not like Forest Mimms's famous book had any shortage of fun illustrations.
Is one supposed to throw a pair of dice or what?
Well, Access is a whole Microshit ecosystem, so if you're looking to develop general RDBMS/SQL skills, it's probably not the most efficient way to go. MySQL and MariaDB are pretty much the same thing, unless they've diverged significantly since the fork. MongoDB isn't an RDBMS, it's a NoSQL document-oriented database. Oracle Express is crippleware, so I don't know why you'd use it unless you have a particular need to get into Oracle stuff. You missed PostgreSQL, btw.
If you're just looking to learn database stuff, SQLite is easy to get started with. Use that. You don't need to worry about the other DBs until performance or huge datasets become an issue.
It doesn't matter if you earn money from it. I don't earn money from programming, but I still do it as a hobby and to fulfill my personal computing needs.
Just start right away with PostgreSQL. It sticks closely to the SQL standard, and doesn't teach you bad habits. It's also no harder to get into than any other DB, because most Linux/BSD provide the necessary packages. There's also very good documentation that comes with it.
Thank you, helpful answers. What about going to PL/SQL later, isn't that an Oracle thing? Does it require an Oracle database server/client?
They even fucked up the equations, left side of Rule 5 should be log a + log b.
Is it better to install the server on a separate machine (or at least a VM)? Heard it's a bad idea to run both a DB server and client on the same system.
Probably, but it's the only officially free Oracle DB, if you want a full version you need to either pay or pirate. And the latest version supposedly supports up to 11 GB of user data, that should be more then plenty for training purposes I guess.
I can understand trying to learn the basics of some stuff via the chinese cartoons route, but employing this to learn math is just a travesty tbh.
oh yeah.
I stopped reading after the first find and dismissed this as a poor quality content, so I didn't notice it.
it looked like a funny and weird idea, but it's crap…
Yeah, but PostgreSQL has its own equivalent (PL/PgSQL). I don't know the differences, but it's the same concept. I wrote some PL/PgSQL (and also the C and Perl variants) for some work-related stuff a long time ago, and it was pretty easy except for the C variant because they have their own memory management stuff instead of libc's malloc, and some of that was pretty wonky (and not documented very well). But if you're just doing the normal PL/PgSQL you don't have to worry about that.
It doesn't matter where you run those, because you're just doing practice stuff and not setting up a big production system that needs to serve tons of requests/second. But if you happen to just have an ARM SBC laying around doing nothing, then this is the perfect opportunity to use it. Otherwise a VM works just as well, except you have to remember to start it up.
The first SQL statement commonly introduced is the SELECT statement, which is only useful if you have some preexisting database. If you have none and have to start from scratch, tutorials which start with SELECT, JOIN etc. are next to useless.
god bless you user
you need to go back to reddit
Is it ok to install Oracle database and MS SQL Server both on one system and have their services run at the same time, or would they conflict with each other somehow?
Kinda further supports his point that it ain't simple, doesn't it.
I've noticed that many database course intructors or professors seem to be women - is there also a large number of women in database tech in the real world (db designers, admins, support etc.)? Does database tech hold much esteem in the general tech industry, or is it mostly considered "spreadsheets on steroids"?
Another question - why is database terminology so inconsistent in both what names are used to label things, and what a given name even really means? For instance, scholars who study the mathematical foundations of databases insist that a table itself is a relation, why anyone whe actually works with databases seems to claim that a relation is how different tables in a database relate to each other (yet another point of view is that one table can represent the mutual relationship of two or more other tables). Another example is one-to-many-relationship and many-to-many-relationship, where "many" seems to mean something different respectively: in one-to-many "many" means "each primary key of table A can appear MANY times in table B" which is completely fine if table A only has just ONE record, but then again in many-to-many "many" supposedly means "MANY different primary keys from table A featured as foreign keys in table B can be matched with MANY different primary keys from table C featured as foreign keys in table B", which seems to imply that tables A and C must have more than one record.
*while anyone who
Guia Mangá Circuitos Eletrônicos is the only way it was distributed. Did a little bit of reading and the US distribution was cancelled for reasons. Only HUEland got it.
What reasons in particular? Would such a manga guide be a threat to homeland security? Or what?
I just found out from a fucking review. Apparently nostarch said no to it. I don't know why. Could be a quality issue. Could be a content issue. Could be the publisher being a dick.
Interesting. So where was it realeased besides Japan, in Spanish-speaking countries?
electron holes hentai
where do you think you are?
Sounds like something that needs to exist.
Seems to only have been released in Brazil, but don't quote me on that.
This could be a good banner
Great, now I also found out that different sources define differently what needs to be done to progress to each successive normal form. The end result after 3NF is kinda the same no matter which source you listen to, but the approaches at each step being different makes the whole thing seem cloudy and confusing. Whay all those inconsistencies?
That looks like cancer.
home of the most severely autistic people on this side of planet earth (besides masterchan)
actually, other side too, we have eurofags
Looks like they're going all-out and don't shy away from cringe anymore, anything goes if it sells. "Hacking iPhones with Kali for Kids" when?
It's really easy to write a shell script that creates and fills table with random data or stuff from text files or whatever. Every DB I used has some kind of shell tool that can do this. Just look at the manual for your DB and you'll probably find examples.
You can import csv into tables, but you cannot "automagically" create the proper relations between this table. Starting out an SQL tutorial with commands like SELECT and JOIN is utterly pointless if no sample database is provided, and unfortunately that's what the manga guide does.
Those relationships are simply foreign key -> primary key mappings. You can generate those yourself in a script. You can even do it randomly if you want (a row's FK = random number from 1 to size of other table's dataset).
People who can generate databases by various methods most likely don't need a SQL tutorial in the first place. What you're saying is similar to a C tutorial that starts off with running a "hello world" program binary through a debugger even though you don't have a binary one and were told nothing about how to write one.
I have the suspicion that most computer science people are highly-functioning autists. There has to be an explanation why the fuck it is so hard to find a computer scientist (or someone working in that field) who can actually explain stuff. Starting an SQL tutorial with SELECT is just as said. I know SELECT is among the more frequently used commands, but it is not the first query in the life of a database.
I learned with the Oreilly MSQL/MySQL book they published in the late 90's. They didn't give me any data set. Reading the chapters on theory plus the stupid examples (make a CD collection database or whatever) was enough to understand how it all fit together. It's not nearly as complicated as writing a C compiler or reverse-engineering a binary. I think that if the book can't explain basic SQL well enough that the reader can make his own tables to play with, then it's just not a good book.
I luv u
The point is that the book progresses from conceptually constructing a would-be-database (with the tables being on paper, in a text editor, a spreadsheet etc.) straight to doing SQL queries such as SELECT on an actual database which the reader obviously doesn't have by that point. They should have either structured the book so that the reader could construct the database first by following the tutorial and only then try to retrieve data from it, or the publisher could have provided a sample database in a handful popular free formats (such as MySQL or SQLite) as a download. As it is, the reader either has to skip ahead on their own to somehow construct an actual database if they want to play around with the SQL examples the book gives, or just keep nodding their head and wait until later.
Is it safe to say that a table basically is a struct holding an array of n (number of columns) pointers to char (the header with the column names) and an array of m (number of records in the table) structs holding n members of whatever types the data in the given columns are?
typedef struct { unsigned int id; char *name; int value;} record;typedef struct { char *header[] = {"id", "name", "value"}; record data[] = {{1001, "fizz", 3}, {1002, "buzz", 5}, {1003, "fizzbuzz", 15}};} table;
Why would that be stupid? Technically it's a suitable subject for a database, and everyone had a collection of CDs back then so it would be engaging to make a database of their own collection.
マンガでわかる電子回路 田中 賢一
It was released in moon as were the rest. Good to know.
Oh crap. Apparently subconciously tried to do two things at once and tried to define struct member values in a typedef (which makes no sense whatsoever). Corrected below in a complete program that creates and then outputs a table.
#include #define COLS 3#define ROWS 3#define COLWIDTH "10"int main(void){ typedef struct { unsigned int id; char *name; int value; } record_t; typedef struct { char *header[COLS]; record_t data[ROWS]; } table_t; table_t foo = { {"id", "name", "value"}, { {1001, "fizz", 3}, {1002, "buzz", 5}, {1003, "fizzbuzz", 15} } }; int i; /* output table header */ for (i = 0; i < COLS; ++i) { printf("%" COLWIDTH "s", foo.header[i]); } printf("\n"); /* output table data */ for (i = 0; i < ROWS; ++i) { printf("%" COLWIDTH "u", foo.data[i].id); printf("%" COLWIDTH "s", foo.data[i].name); printf("%" COLWIDTH "d\n", foo.data[i].value); } return 0;}
Fan translation when?
Is it a reasonable thing to do to split a formatting string like that?
Yes, it's the standard convention for using macros in string constants.
How would you go about making the numbers in format specifiers variable at runtime? Or do they need to be constant at runtime?
yea, no
You just edit a string and then use that string which you just modified as your format string. Keep in mind that if the user is able to control the format string that you might have a format string vulnerability.
I don't think that things such as format strings should be modified directly based on user input. Much rather program logic should modify them in a well-defined and predictable manner only after user input has been parsed (assuming user input may have relevant influence).
There's nothing arcane about the psql or mysql CLI tools, unless maybe you don't understand the very basics of Unix shell.
Database thread:
oh seriously