Ventilate Fedora Anaconda by moving software selection to firstboot step

Currently we have anaconda for Fedora 18 designed as a hub with spokes. One of the spokes is the software selection task.

There
are a few reasons why it may be desireable to move it from anaconda to First boot.

a) Simplify anaconda functionality
b) Reduce the livecd size so that it is under 700megs
c) For smaller systems (netbooks, or systems with 1 gig of memory), allow the firstboot to specify a larger set of installable software. (Eliminate the mutually exclusive selection imposed by Fedora 18).
d) One gig systems do not have sufficient memory to install every option within the Gnome choice. Ditto for some of the other DEs.
e) If software installation is moved to firstboot, Fedora is already booted, and firstboot software installation (yum or whatever) can take advantage of the
/home/user folders increased disk space availability.

The above are just my thoughts after having to do the fedora installation twice because of d)

I am not a Fedora internals person and I really do not know if my idea is practical. It may not be if the livecd option is one reason for not considering this idea, there may be others.

Quote:

---

Originally Posted by lsatenstein

Currently we have anaconda for Fedora 18 designed as a hub with spokes. One of the spokes is the software selection task.

There
are a few reasons why it may be desireable to move it from anaconda to First boot.

a) Simplify anaconda functionality
b) Reduce the livecd size so that it is under 700megs
c) For smaller systems (netbooks, or systems with 1 gig of memory), allow the firstboot to specify a larger set of installable software. (Eliminate the mutually exclusive selection imposed by Fedora 18).
d) One gig systems do not have sufficient memory to install every option within the Gnome choice. Ditto for some of the other DEs.
e) If software installation is moved to firstboot, Fedora is already booted, and firstboot software installation (yum or whatever) can take advantage of the
/home/user folders increased disk space availability.

The above are just my thoughts after having to do the fedora installation twice because of d)

I am not a Fedora internals person and I really do not know if my idea is practical. It may not be if the livecd option is one reason for not considering this idea, there may be others.

---

That's a pretty lofty change proposal considering that the advantages only benefit systems that are long past their prime and shrinking in popularity. I know there is an appeal to keep old outdated hardware going for some but keeping old limited hardware going is actually a hindrance to forward progress in development.

Quote:

---

Originally Posted by deanjo

That's a pretty lofty change proposal considering that the advantages only benefit systems that are long past their prime and shrinking in popularity. I know there is an appeal to keep old outdated hardware going for some but keeping old limited hardware going is actually a hindrance to forward progress in development.

---

===
When you quote old technology, realize that while two of my three systems are 4 gigs or larger, I also have netbooks that were manufactured by Intel in 2011. That is not old equipement that is past their prime. There are many netbooks made for Asia, Africa, Latin America, and elsewhere that are with 160 to 256gig drives but with 1 gig total memory. These netbooks were designed to be rugged and for students.

Fedora18 runs for about 4.5 hours non-stop on them. I have two of them because I write software. I write for French and Spanish users.
Currently my software is designed for 32bit and 64 bit hardware. (Different compilers and libraries). When you realize that on a two user system, you may have one user who wants KDE and the other Gnome, then you really want to install both in one go. To keep the installation time reasonable, Fedora 18 limits you to installing one and only one DE.

Please also note. Ubuntu is already on tablets. Most tablets have 1 gig memory and up to 32gigs flash. What I propose is to ventilate anaconda so it is easier to support and to allow a livecd image to fit on a 1 gig flash drive.

Have you ever had to restart an installation because something was configured wrong, and you have to wait for firstboot? If you have, then you may find merit in my asking if it is worth researching to evaluate the pros and the cons of this change. The pros and cons include development and support costs, program sizes, modularity, development and quality control activities, security and anything else I forgot to mention, if installing from dvd image on flash, instead of taking 40 minutes, the system can be at firstboot in 10, and that means, the installation was successful. We move about 35 minutes of elapsed execution from anaconda and add it to firstboot. If from first boot, only the latest software gets installed. now, my experience is that the snapshot prepared for the DVD is three months behind in time from the actual go live date. We do the installation, we do the first boot, and we run yum to download 300megs of current updates. By relocating the software installation to immediately follow firstboot, a lot of elapsed time is to be saved.

We must not put sticks in the wheels (batons dans les roules). If the Fedora architects think there is merit, they will build a comparison matrix of pros and cons, and draw a quantitive conclusion.

Quote:

---

Originally Posted by lsatenstein

===
When you quote old technology, realize that while two of my three systems are 4 gigs or larger, I also have netbooks that were manufactured by Intel in 2011. That is not old equipement that is past their prime. There are many netbooks made for Asia, Africa, Latin America, and elsewhere that are with 160 to 256gig drives but with 1 gig total memory. These netbooks were designed to be rugged and for students.

---

Netbooks as a whole are past their prime. The entire industry is abandoning them from the manufacturers to even the distros that were optimized for netbooks are disappearing as well. Most of those past developers have jumped over to putting their efforts into ARM work now. In short it is a lot of extra effort to do what you propose that will continue to yield diminishing returns for that effort.

Netbooks are not obsolete. Dual core 64 bit atoms with very low battery power consumption, with keyboard, usb ports and the like, and with a $150ea cost are most certainly attractive for schools where the government provides the device. Some Netbooks include a TPM,

The TPM has an integration with the motherboard to permit killing the netbook if the item is reported stolen. The TPM is used to hold licensing and password information, and for college software engineering students, a random number generator that generates numbers uniformly distributed.

Many governments are delighted that they can provide these inexpensive units to students. Many students live in remote locations, where only internet access is the government or school buildings. Yes, these backwoods areas are old-fashioned and obsolete. What is wonderful, is that unlike a tablet which costs 300+ and has at most 32gig free space, the minimum free space on these units is 250gigs. Enough space for all electronic copies of their text books and classroom assignments.

You would not necessarily put a programmers workbench as a main application, but it is a consideration, given the tablet alternative cost.

Obsolete is very much the wrong word. Well matched technically and economically to the need is what I am describing.

You must live in a country where wifi and Cellphone technology is low cost. Not so in the countries where these devices are targeted for distribution. Fedora 17 and Fedora 18 work just fine on these units.