| Thread Previous • Date Previous • Date Next • Thread Next |
Hi Nio,This is one of the odd things about using the -9 compression, at least I think The entire size should be roughly the exact same. However there is something that changed enough to make the compression algorithm compress things slightly different and voila 25MiB more :(
On a side note, there may be something that slipped in somewhere, but I do not see it. I have had these issues before where running the compression a few times tweaking minor things produces a huge difference (685 to 725 is huge)
But really in precise it wasn't a huge deal. systemd is a huge deal and has contributed to this extremely large footprint.
I definitely have considered the 2 file method, and have thought to work on dltbl some to make it more robust. The main issue is that we need to do a few things. 1) get an accurate list of the tarballs at the time the ISO is built, and have dltbl get an accurate list when run. 2)Automatically assume a sane default (even if things change... always look for the best tarball)
3)Test installing from a USB tarball pretty heavily4)Make the automatic tarball selection code look on USB devices to automatically choose a good tarball if there is not one on the ISO 5)update phill's server with a good selection of tarballs, and do things nicely (offer a 64bit tarball, tarballs with a bunch of preinstalled applications, etc...)
I do not think I will strip the system more, I will purge my cached debs, and rebuild once I make sure things work correctly. This usually corrects the issue (not sure why....) It may be that the chroot updater brings in a few unwanted extras when I run it in the mkISO or UpdateTarball scripts.
Anyhow this whole process is much easier now that I don't have to do it all in a VM, so once I find the issue I can rebuild it more quickly.
On 07/29/2016 12:55 PM, Nio Wiklund wrote:
Hi Israel,I recognize the sleep fix - I use it in mkusb to make things work in a reliable way :-)But that should not make the code grow to 725 MiB. Did you add code, or are there updates from upstream (packages that grow outside your control)?I know that Lubuntu is suffering from such problems, not only the kernel but also the language packs (including some Asian locales added big chunks) and other software grow.-o-Have you considered the 'two file method' with an iso file and a separate tarball? The tarball can be written to another drive, for example a USB drive or a second CD (a data CD), and accessed after booting from the CD.The OBI can also download a tarball when it is running, but I think some users will prefer to get it separately. (The tarball can be stored in RAM, but I think the old computers, that cannot boot from DVD or USB also have too small RAM to store a tarball.)The 'two file method' might be a solution to the problem with growing software components and the CD size limit. Otherwise I'm afraid that you must strip the system too much so that it will lack important features.Best regards Nio Den 2016-07-29 kl. 19:30, skrev Israel:Thanks guys! So here is the information i have found so far with the menu bug. I have been testing things quite a bit. There is some issue with how all the startup commands are handled that was causing a mess, as too many processes were started too closely together, OR there was a problem with too many files created simultaneously OR too many pipes needed. kinda broad, I know... However Joe (The 'J' in JWM) has been giving me some things to try Here is the link https://github.com/joewing/jwm/issues/321 I tried recompiling things with sleep(2); in a few places to see if I could 'make' it load correctly However, I have made some tweaks (using the mkISO/UpdateTarball scripts) which now has a working liveOS, and (hopefully) installed OS. The issue now is that the size is 725MiB If anyone uses the ISOmaker scripts, please do a git pull I recently updated the help screens (using the arguments -h|--help|-?). The ISOmaker help screen now covers most everything the program does, and the ways you can modify things. I also added more options for customization, and added info on what the log file is called. I also updated the mkISO/UpdateTarball scripts to allow custom jwmrc files (which is how I have been testing things) Just run on of the programs with --help (or -h or -? if you don't like typing long words) to see the information. Please let me know if anything is still not very clear. I have not pushed any changes to the settings manager or torios branches yet, as I still have a lot of work to do on settings manager, and need to verify these things FULLY fix our issues before rebuilding torios packages. On 07/27/2016 05:49 PM, Nio Wiklund wrote:+1 :-) Den 2016-07-28 kl. 00:28, skrev Cinque Port Computers:Hi Israel, Thanks for the update and nice to know you're engaging in real life pursuits too :) Cheers, JackT ..
-- Regards
| Thread Previous • Date Previous • Date Next • Thread Next |
