How to optimize building from source.

Hi, ive read about MAKE options to optimize the package youre building, like "-march, -O3, -pipe and -fomit-frame-pointer"
Is there another ways to improve package speed and optimize?

bangkok_manouel wrote:
I'm surprised no one's using the new graphite branch loop optimizations:
-floop-interchange -floop-strip-mine -floop-block (See http://gcc.gnu.org/gcc-4.4/changes.html)
Using this since GCC4.4 was pushed in [TESTING], no build problems at all. Of course, the performance gain is >>!*HUGE*!<< but won't be noticed by any human being. ;-)
Found this in another thread... not sure how sarcastic he is being... is it worth using those optimizations?

Similar Messages

Building from source -- how to remove installation?

Greetings.
I am relatively new to the world of Linux, and while google has helped me thus far, I can't seem to solve this one: How do I effectively keep record of all the files being installed, when building from source?
When everything is done through Pacman or the AUR, it's all in a system; however, not when building it manually..
Some sourcefiles has an "make uninstall"-function, but not all.
I kinda feel that I don't have control over what's happening on my system, how do I solve this?
Thanks.

I only recently learned how to create my own packages. It was so easy! You can read about it here: http://wiki.archlinux.org/index.php/Building_Packages That link also provides a simple example "PKGBUILD" file.
As you can see, the "build" function in the PKGBUILD file is nothing more than what you would do to build the package by yourself. But, by creating a PKGBUILD file, using the "makepkg" command, and then installing it with "pacman -U package.pkg.tar.gz", you can easily uninstall it in the future.
If you want, you can then upload your PKGBUILD file to the AUR, and feel good as an Arch Linux contributor.
I found it interesting that Arch Linux doesn't even setup "/usr/local/bin" in the PATH by default. It's probably for the best. It keeps me from being naughty.

[SOLVED] Building from source?

I'm new - quite new to Linux, totally new to Arch - thus this newbie question.
I can't seem to understand how to properly compile from source in Arch.
And before you ask, I have read the Wiki. A lot.
The way I understand it, normally one would use pacman and rely on it to find, download and install the package. I have not yet got a proper internet connection set up, and will therefore have to download the package(s) from my Win XP computer I'm sitting at.
I want to download wvdial - for getting dialup up and running. wvdial seems to depend on ppp, wvstreams and xplc, so I have to download those as well.
After searching, I found them all, but only the source code, as gzipped tarballs.
I understood that one was then supposed to use makepkg to make a package from the source code, and then install the resulting package using pacman.
Sounds OK, but it looks like I need a so-called PKGBUILD file for each package, a script used by makepkg to build the package correctly.
Did those tarballs contain any PKBUILD files? No. There were makefiles, but no such thing as a PKGBUILD file.
What am I supposed to do, then? Did I get the wrong tarballs?
I got wvdial from this page: http://www.archlinux.org/packages/extra/i686/wvdial/ which took me here: http://alumnit.ca/wiki/index.php?page=WvDial which again lead me here: http://alumnit.ca/wiki/index.php?page=DownloadReleases
This last page also included wvstreams.
For ppp: http://www.archlinux.org/packages/core/i686/ppp/
For xplc: http://www.archlinux.org/packages/extra/i686/xplc/
Last edited by Mzg (2009-09-06 18:13:28)

the binary package have to be downloaded from a mirror. here is a list with them: http://wiki.archlinux.org/index.php/Mirrors
note: http://www.archlinux.org/packages/extra/i686/wvdial
looking at the link you can find out where to look. is in extra repository, i686
http://linkmirror/extra/os/i686/

Wombat (Arch 0.7) Kernal build from source problem

Using the large Arch 0.7 ISO and selecting build Kernel 2.6 from source, the command "make menuconfig" fails because the ncurses library is not present.
I am able to start the build of the kernel using the default .config file. There were a number of warnings for configuration variables no longer being used, and a number of features had been deprecated. Its not finished yet, but I suspect it will work ok.
I may install the pre-built kernels just in case the config file is incorrect.
Should I put this in as a problem, or is it already fixed in 0.8?

Actually, the instructions are quite good and I did follow them
It's pretty easy, since you can alt between F1 and F2 and check what you have to type.
I going through the whole process again, since I did not like the error messages I was getting. As in the earlier case, I am only installing base (which includes ncurses.)
I am not going to build the kernel from source this time, but wait until the system is up before doing that.
I have never seen this problem before with the beta's of 0.7, so I am going to assume its a fluke. I have used beta's of Wombat on 9 machines and this was the first time it happened.
If no one else has had this problem, I am going to assume its my fault.
No big deal.

How to read ZIP from Source System

Hi frnds,
In my sender side i have Zip file, I need to read from source directory, File adapter will provide option to read Zip file??
if i want to generate out put in ZIP format is it possible??
tell me what are necessary steps required??
Regards
Raj

Hi Raja,
All the steps will be same as a file to file scenario. Nothing new except that you need to configure your sender channel as mentioned in the blog that i asked you to refer. Then you just put your file with extension .zip in the source directory. This is in case you dont need to process data(no mapping required).
In case you do not need to create a mapping(just transfer data from one place to another), you can directly configure your scenario. No need to do anything in the IR.
Hope it makes it clearer.
Thanks and Regards,
Sanjeev.

Building from source

Hello,
I need the application Xwisp2 from http://www.robh.nl/picsoft.php. This software isn't downloadable as binary but only as source. I'm very unknown to programming, building a package with a makefile I can, but this even doesnt have one. Can anyone explane me how I turn this into an excecutable application please.
I'd be very happy.
Thank you
Mac OS X (10.4.8)

The answer depends on the number of files that are part of the distribution and the language in which they were written. In the simple case, if you have a single source file, you can typically build it with make without a makefile. For example, you can generally build foo from foo.cpp with: make foo

Haguichi won't build from AUR, and won't build from source tarball.

I managed to get Hamachi installed. Go me. But I like having a frontend for things like that, so I tried to build Haguichi from the AUR. Didn't work. I downloaded the source tarball, and running ./configure returned the same error as yaourt.
redpill% ./configure
checking for a BSD-compatible install... /usr/bin/install -c
checking whether build environment is sane... yes
checking for a thread-safe mkdir -p... /usr/bin/mkdir -p
checking for gawk... gawk
checking whether make sets $(MAKE)... yes
checking whether to enable maintainer-specific portions of Makefiles... no
checking for pkg-config... /usr/bin/pkg-config
checking for dmcs... /usr/bin/dmcs
checking pkg-config is at least version 0.9.0... yes
checking for GTK_SHARP_20... yes
checking for NOTIFY_SHARP... yes
checking for GCONF_SHARP_20... yes
checking for GLIB_SHARP_20... yes
checking for DBUS_SHARP_10... no
configure: error: Package requirements (dbus-sharp-1.0) were not met:
No package 'dbus-sharp-1.0' found
Consider adjusting the PKG_CONFIG_PATH environment variable if you
installed software in a non-standard prefix.
Alternatively, you may set the environment variables DBUS_SHARP_10_CFLAGS
and DBUS_SHARP_10_LIBS to avoid the need to call pkg-config.
See the pkg-config man page for more details.
I made sure the dependencies were installed and updated to the latest versions, but it still won't build. What am I doing wrong?

Scimmia wrote:Did you read the comments on the AUR page?
Just did. I guess I'm still a bit noobish

MySQL custom build from sources best practices advices

Hi,
These days I'm setting up something that could be called a home server, basically a web server with DB support.
Of course it's based on Arch Linux (probably not the best choice while everyone around is speaking about CentOS, RHLE, Debian, etc as server-oriented linuxes, anyway I hope Arch is not worse).
Every application I need is available in packages, but the idea is to build every app to a custom and better organized location.
I'm not a fan of the applications' distribuition in /bin /sbin /lib /var /etc ...
I embrace the philosophy of "one application in one place" with a structure similar to
/appdir
|---/app1
|---/bin
|---/lib
|---/conf
|---/docs
|---/logs
|---/tmp
that's why I would like to build all my applications using the same pattern.
Now I want to build the MySQL database server, but the process it's not straightforward like for other applications.
I use the following configure inspired from --help, MySQL docs and PKGBUILD provided by the Arch mysql package:
CFLAGS="-fPIC -march=amdfam10 -mtune=amdfam10 -O2 -pipe -fno-strict-aliasing -DBIG_JOINS=1 -fomit-frame-pointer" \
CXXFLAGS="-fPIC -march=amdfam10 -mtune=amdfam10 -O2 -pipe -fno-strict-aliasing -DBIG_JOINS=1 -felide-constructors -fno-exceptions -fno-rtti" \
./configure \
--sysconfdir=/usr/local/apps/mysql/conf \
--prefix=/usr/local/apps/mysql \
--exec-prefix=/usr/local/apps/mysql \
--sbindir=/usr/local/apps/mysql/sbin \
--docdir=/usr/local/apps/mysql/docs \
--mandir=/usr/local/apps/mysql/man \
--datarootdir=/usr/local/apps/mysql/share \
--libexecdir=/usr/local/apps/mysql/sbin \
--localstatedir=/usr/local/apps/mysql/data \
--with-mysqld-ldflags=-all-static \
--with-zlib-dir=bundled \
--enable-assembler \
--without-debug \
--without-docs \
--without-readline \
--with-mysqld-user=mysql \
--with-libwrap \
--with-charset=utf8 \
--with-collation=utf8_general_ci \
--with-extra-charsets=binary,ascii,cp1250,cp1251,koi8r,latin1,latin2,latin5,utf8 \
--with-embedded-server \
--with-unix-socket-path=/var/run/mysqld.sock \
--enable-local-infile \
--with-plugins=partition,ftexample,blackhole,csv,heap,innobase,innodb_plugin
Could someone with a deeper understanding of the process to give me some advices about optimization, plugins, etc, please?
I want to build a fast server without unnecessary stuff.
Some questions are about paths and directories:
1) is it possible to get the pattern described above without having paths like mysql/mysql/..., mysql/lib/mysql/<files>, mysql/include/mysql/<files> but mysql/lib/<files>, etc ?
2) is it OK to delete mysql-test and sql-bench directories if I don't need them ?

bigman2126 wrote:
Stebalien wrote:2. Never build wireless/wired modules into the kernel. If you use fuse, don't build that into your kernel either.
I would like to know the reason too, I tend to build my custom kernels so that the hardware and things I use are compiled into the kernel.
bangkok_manouel wrote:because the firmware is loaded later so the module is not properly initialized (or some word that sounds like this)
That *has* to depend on the specific wifi driver because I have iwlwifi-1000 built into my kernel as we speak and it loads the firmware fine.
As far as i915 and modesetting goes, I had the exact same issues. It turned out in my case another driver was stealing control before the intel kms one could grab it (maybe FB_VESA?). Anyways, here's the relevant part of my kernel config that does work with i915 kms built-in:
http://pastebin.com/jaCAqPjh

Google Chromium takes up to 10 gigs to build from source?

So I was browsing the AUR and saw in the package build for chromium-browser-svn that the build can take up to 5gigs to build. Well after surfing the web the official chromium wiki states that it can take up to 10gigs.
This seems crazy to me, though I don't have a lot of programming knowledge.
So my question is, does anyone here know why a program like this, that doesn't have a very large binary, would take so much disk space to build?

Wow, I didn't realize that the source was 900mb. But still, 10gb is a whole order of magnitude greater than the source. I don't know, maybe I just don't know enough, or anything for that mater, about how software compilation happens, but this still seems quite ludicrous to me.
So is there a simple answer to this or is it quite complicated?
Thanks for any helpful information.

How to Optimize 3D from Mac to 3D HDTV with Apple TV

I have been trying to stream a 3D movie from my Macbook Pro to my 42" LG 3D 1080 HDTV via Apple TV, and it works, but not perfectly.
The details: I have an MP4 formatted 3D movie, which plays with horizontal side-by-side frames. Running the movie in Quicktime at full screen, I mirrored my Macbook to my TV via Apple TV and then turned on the 3D feature on my TV. The TV converted the side-by-side frames to 3D and it appears in 3D when I wear the 3D glasses (they are passive glasses), but the image I see is not quite perfectly in focus, it seems off in regards to clarity, and there is a distracting vertical bar appearing on the right side of the screen (some sort of remnant cut off from the main picture, which looks like a reflection or something from the video).
What to do? I tried to adjust the very limited 3D settings the TV offers, but it had no effect. I know the 3D feature on the TV works correctly because I've tested it with ESPN 3D content streamed from my cable box and it looks really good.
Any suggestions? Are there any other video playing apps that could optimize this process?
I was thinking of copying the MP4 file to a memory stick and just plugging that into the TV, but I really wanted to see if I could get this to work from my computer so I could have access to a library of movies with more ease. I like the concept of High Definition 3D but I do not have the desire to invest in a Blu-ray 3D player right now, so I just want to make this work with my existing hardware.
Thanks for any suggestions and help.

A quick update on the situation, I solved a bit of my problem, but not exactly as I had hoped; by removing my Macbook and Apple TV from the equation I successfully played the 3D movie on my TV.
The solution to get the movie to play correctly was to load the MP4 onto a USB thumb drive and plug it directly into the TV. Movie played, 3D worked (better, without artifacts and imbalances) and I was even able to turn off the subtitles through the TV's settings, however, my attempt to wirelessly mirror the movie from Macbook to TV via Apple TV was a fail.
2 more things: Although the 3D "worked" it still didn't feel as robust or have the depth I experienced when viewing ESPN 3D, and after 10 minutes of watching the film in 3D my eyes started to feel funny (first time this ever happened with passive glasses), so I switched to 2D mode.

How to optimize MF Media Source for MKV playback (containing H264 and AC3)

I've built a working Media Foundation Media Source for matroska (MKV files) using the sample for an MPEG Source. I've finally gotten it to a point where I can play the file and both video and audio are visible and audible. I followed the MPEG1Source
code quite closely actually.
However, I can't seem to make the video playback smoothly. Audio plays fine, but video just can't seem to keep up with the playback speed. It stutters and breaks up into blocks constantly.
The only thing I have been able to try is to remove as much dynamic allocation of memory as I can. While my parser parses the file and initially creates a bunch of objects on the heap (pertaining to track formatting, cue locations, etc)
, when I start actually parsing video and audio frames, I'm no longer dynamically allocating memory on the heap. I even removed a std::queue for frame lengths and created a fixed length circular buffer instead.
What else do I need to try?
I am simply parsing each of the video and audio frames in the order that they appear in the file. The frames are not Annex B compatible (for H264), so in the DeliverPayload function I replace the length section of each NAL unit within the frame
with a start code of 00 00 00 01. Then I copy the frame into the stream's buffer.
One thing that is very different from the MPEG1Source example is that I am not creating structs that contain all of the video "initialization" data (i.e. MPEG1StreamHeader, MPEG1SystemHeader, MPEG1PacketHeader). Instead, I am leaving all
of it as public variables on my parser class. So I am referring back to my parser quite a bit from the MKVSource class. I don't know if this affects anything.
Anyways, I'd be happy to elaborate to get this working... Thanks!
Lee McPherson

Just a follow-up:
I was indeed calculating the timecode / timestamp incorrectly. MKV uses a technique called lacing to stitch frames together and save space. While I was separating them and forwarding them correctly to the pipeline, the timecode in the header
for each laced block only applies to the first frame. I needed to add the default duration time to each successive frame laced in the block.
While not specifying the timestamp using the SetSampleTime method caused smooth video, it was actually playing faster than the audio. Each frame _needs_ a timestamp to sync correctly with the audio or other streams.
Lee McPherson

HOW to install fonts from rpm/tar.gz (ec-fonts-mftraced)

Hey all.
I'm trying to install a music program called Lilypond, for my girlfriend, as an example of how easy (heh) linux is to use.
Now, to use this particular version, I need to install some fonts. ec-fonts-mftraced, to be exact.
The Lilypond site provides downloads for the fonts here
There is an rpm, deb, and tar.gz version, none of which I've managed to install from. I've tried using
rpmunpack | gzip -d | cpio -uid
as (kind of) per the instructions:
EC-FONTS-MFTRACED
GNU LilyPond 2.3.x needs TeX (if you are using the TeX backend) and
PostScript Type1 PFA/PFB fonts with full internationalisation support.
By default, LilyPond uses the ec* family.
INSTALLING EC FONTS
The binany RPM and DEB packages should be installable on any RPM or
DEB based system.
If you do not have such a system, install either
* the RPM, like so
rpm2cpio ec-fonts-mftraced-1.0.2-1.noarch.rpm | cpio -uid
mktexlsr
* or the DEB,
ar p ec-fonts-mftraced_1.0.2-1_all.deb data.tar.gz | tar zxf -
mktexlsr
BUILDING FROM SOURCE
cvs -d :ext:[email protected]:/cvsroot/lilypond co sauter-fonts-mtraced
cd sauter-fonts-mtraced
make all install
mktexlsr
I can't get anything to work though, nor can I install the fonts thru KDE control centre.
Can anyone help me with this problem?
Much appreciated,
Thom.

You need t1utils (an easy compile), potrace (or autotrace), tetex, and python so you can compile mftrace.
mftrace PKGBUILD is at the bottom of http://bbs.archlinux.org/viewtopic.php?t=6450
Then you need to:
cd /usr/share/mftrace/
import filelib
Here's the PKGBUILD for ec-fonts-mftraced:
http://bbs.archlinux.org/viewtopic.php?t=6451
Fontforge I think helps it compile faster (??)
All of the above-mentioned packages that aren't available via pacman have been uploaded to ftp.archlinux.org/incoming

How to build a package from sources

hi all,
Can someone please give some advices regarding making binaries from sources. I have found that a lot of programs on Snow Leopard can be also run in terminal and are in source form and need to be build. I have installed Xcode already and I have some programming skills as well such as using MinGW (GNU GCC) and VS EE on WinXP. I have never done anything in any UNIX-like system and I know sources need to be just compiled.
Please, give me a link if you have any =)
Regards.

I think you have done something in a UNIX-like system. MinGW is designed to provide a UNIX-line build system with GCC for Windows. Now, however, you have the real thing.
In most cases, all you need to do is extract the source tarbar into a directory (or checkout via CVS or SVN). Go into that directory with the Terminal and type "configure" and, if that goes well, "make", and if that looks good "sudo make install".

Building eveything from source with ABS

Hi. Im amazed with the ABS system. It´s aewsome, since I love to build all the software in my system from source, and compiling it and make it a binary package is just genious. But I have some questions, here they go:
1 - Im compiling everything using the ABS tree using makepkg -b, so it compiles and generates packages also for its dependencies and I edited makepkg.conf so all the generated packages go to a folder (i.e. /packages) so I can install them later using pacman -U package. My question is: If I later install a given packages, that depends on another, will that package be automatically installed when it´s needed? And is there a way to automatically redo all the source compiling and package generation automatically upon a source update for the packages I previously generated (like syncing in pacman)?
2- Im trying to build my desktop from source the ABS way the cleanest way possible, without all that kde programs, just the core of its window manager, so I would like to know what sources to compile. I guess it would be something like xorg -> kde -> nvidia drivers -> compiz-fusion (not supported by ABS) . Any directions, guys? Thanks in advance.
Last edited by Arthanis (2008-01-28 00:32:24)

Arthanis wrote:Sorry, it´s because i didn´t know how to move the post, I saw that if fit in another part of the forum
Don't worry too much. I suppose I was being mean given you are new here. It is just that help tends to be better when clustered in one place.
jacko wrote:I thought about doing this for myself at some point, there is a lot of things I just do not need that is compiled in by default which I could rid myself of but, is all the time worth it in the end? Would like to hear someone elses expertise on this issue.
Well, you could also add in more optimization for your specific processor. But I honestly don't think you need this for every package. Perhaps the kernel, glic and a couple of other often used and big packages, but in the end it is barely noticeable given Arch is fairly optimized by default and can make you system less stable.
The only real reason to recompile is to add/remove options. If the option is easy to add and doesn't bloat the package. the devs will probably add it for you if you ask in the bug tracker anyway. I think the only package I now recompile are mozilla ones because I want the branding.

How to optimize text editing performance in Flash Builder 4?

I am running Flash Builder 4.0.0 (build 272416) on a Macbook Pro 2.4 Ghz with 4Gb RAM. I am experiencing extremely poor performance while simply editing moderately complex MXML files.
By poor performance I mean noticeable lag between keystrokes. Type one char, wait 2 or 3 seconds on a 700 lines MXML with about 20 buttons, 2 AdvancedDataGrids, 1 text area. Fairly complex, but that's what I need.
Opening said MXML file takes around 12 seconds.
Saving said MXML results in around 8 seconds: no automatic building, nothing else being done.
On trivial MXMLs there's still a noticeable lag while typing, around 0.5 seconds per char.
At this point Flash Builder 4 is unusable for me: editing .as files is still decent, but for MXMLs is almost impossible to work with, as you can see from the numbers above.
Is there *anything* I can do to improve text editing performance? I've disabled all sort of niceties (automatic completion, indentation and so on) in the prefs, sadly to no avail. This is a major inconvenience for us.

You understood correctly:
- I removed all Build Path > Source Paths
- I moved those paths over to Library Paths > Source attachment
I took inspiration by how the Flex SDK is added to the project. Flex and flex related frameworks (e.g. datavisualization lib) also specify a Source Attachment under Lib Path. If I remove the source attachment I am not able to drill down into Flex code when I want to by Command-clicking on a class name (or hitting F3).
All my SWC are open source. BTW some of the source paths I was using were the path to the Flex code and to datavisualtization lib, which I think are the biggest library I am linking against. So maybe to recreate, just add those 2 libraries and make sure to add the source path AND source attachment.
I am using RSLs for every library if that matters.
I don't understand why specifying a Source Attachment would cause any issues... the swc is in sync with the source specified in its source attachment.

How to optimize building from source.

Similar Messages

Maybe you are looking for