On Tue, 29 Sep 2015 02:01:03 -0400, Stephen Chrzanowski <pontiac76 at gmail.com> wrote:
> For the past year, I've had a script that runs daily that reads all the > links off of http://sqlite.org/download.html and downloads anything that is > missing. It has been a long while since I've looked at this particular > section of my NAS, but thought I'd bring my repositories up to snuff, and > build my own DLLs against each revision posted again. (So, yes, Richard, > as a prank, you could slap a 1tb garbage file up and my script would > dutifully download it.... My ISP would LOVE more of my money) > > First, I noticed that there hasn't been anything posted new for the > amalgamation a long while (July 30 for 3.08.11.01 according to my file > systems time stamp), so thinking that my script broke for whatever reason, > curiosity bit me. I checked the download site and no, things seem to be > working. So a testament to stability, both for the code, and surprisingly > for my script!! > > Second, I noticed that on the download page itself TODAY has no reference > to files named "sqlite-amalgamation-YYYYMMDDHHmm.zip" yet I have a bunch of > them in my archives. Are these the files put up for the purpose of the > pre-releases of a finalized build and my script is working better than > expected??? (I might get the script to filter those files out, but being a > digital file packrat....) > > Third, I don't know if it is something that can be done now, but I've > noticed that on a very few of these downloads, I'm getting file sizes in > the 5kb range. Looking at the raw bytes of the zip file in a text editor > is a bit strange, but it looks like the .zip file was downloaded as an HTML > file. (As in, rename the zip file to .TXT and open up in notepad, kinda > raw content). I'm thinking that the web server didn't find the file when I > requested so just spit out the page. > > Fourth, a little bit of hand-holding might be needed, but do the zipped > archives of the released amalgamations (Not the dated files) exist on the > web server somewhere, or can I get the zip from the SQlite repo? If only > from the repo, where can I get either the .c/.h or .zip file for older > versions? (Just in case I miss a revision in the future because of that > 5kb thing) -- Maybe because of that 5kb thing I should write into my script > that any zip file that is less than 100kb should be just flat out deleted. > I'm a somewhat interested in getting older versions for the sake of just > having them, and going back to them in case I want/need them. By far the easiest way to follow revisions and build older versions is to follow the repository, which is managed by fossil. http://fossil-scm.org/index.html/doc/trunk/www/quickstart.wiki Clone the repository using fossil: fossil clone http://www.sqlite.org/cgi/src \ ~/var/fossil/repo/sqlite3.fossil Open a checkout in a dedicated directory: mkdir -p ~/src/sqlite3 # once cd ~/src/sqlite3 fossil open ~/var/fossil/repo/sqlite3.fossil trunk Update it periodically: cd ~/src/sqlite3 fossil pull and update the checkout: cd ~/src/sqlite3 fossil update trunk # or any other point on the timeline Building the amalgamation is a matter of: mkdir -p ~/bld/sqlite3 # once cd ~/bld/sqlite3 ./configure [ options ] Then you can i.e. build the command line tool and sqlite3_analyzer: cd ~/bld/sqlite3 make clean make sqlite3.c make tclsqlite3.c gcc ${CFLAGS} -o sqlite3 sqlite3.c \ ../../src/sqlite3/src/shell.c make sqlite3_analyzer Your build can grab the version in the generated sqlite3.h in the build directory: #define SQLITE_VERSION "3.8.12" #define SQLITE_VERSION_NUMBER 3008012 #define SQLITE_SOURCE_ID "2015-09-26 17:44:59 33404b2029120d4aabe1e25d484871810777e934" Hope this helps. -- Regards, Kees Nuyt
