On Fri, May 18, 2018 at 02:51:00PM -0700, Paul Rogers wrote: > > > ... But if you want to go on to build a desktop with a modern > > graphical browser then 8GB RAM (and even with that, maybe swap if you > > are doing other things during the compilation) is more comfortable. > > > > For a server, it probably varies between different packages. Some > > are fairly light to build. > > > > So no, for LFS-only, 6GB should be adequate. And I suspect many > > ... > > Just as a data point, I'm running 7.10, XCFE, Firefox, LibreOffice on an old > Conroe Core2-Duo 6700 (non E-), 4GB. This is my "daily driver". It's "fast > enough", and that's all I need. I don't need a computer so fast it will > finish things before I've got it all thought out, if you know what I mean. > Sometimes it can be a good thing when one can't just "brute force it" and has > to do it with some finesse. > > I'll do building on an old 12GB i7-940, so's I can use -j8. > Since that was a reply to my reply, I'll bite:
I prefer to run current versions of graphical browsers, or their engines (e.g. qtwebewngine for falkon, webkitgtk if somebody uses a browser based on that) because of the many vulnerabilities which eventually become known. My preferred browser for general use is firefox, building recent versions of that with less than 8GB is painful because of the amount of swapping - even if the drive is an SSD. And I used to try to update firefox on previous "released" systems, sort of in a "it can be done" fashion - I used to like the idea of being able to support a system or 3 years, but changes in glibc (fixes of historic vulnerabilities) and then in firefox made that not possible for me. Gradually I have added to my machines, and I realised that for me there was no benefit to updating anything older than the previous release. But after firefox-60 came out I decided it would be nice to be able to update the last-but-two release (8.0) in the spirit of "expecting users to update the whole system more than once a year is a pain for them". So I tried that on my i7 haswell (16GB, SATA SSD), which builds relatively quickly. Updated sqlite, nspr, nss. icu, graphite2, harfbuzz, rustc. But then firefox failed to build one of its rust crates. Adding --verbose to the mach invocation did not give any more information about why it had failed, and anyway failures in packaged rust crates are often terminal - if you try to patch or sed something to fix an error, a hash check will later decide you didn't build the expected version. The point is that for current software, with the continuing changes in C++ and other flavour of the month languages, the more horsepower and memory, the greater the chance that it might build. But of course my machines are more of the 'my lab' flavour - a heterogenous collection, and I expect to build on each of them rather than build on one and then roll out the binaries. ĸen -- This email was written using 100% recycled letters. -- http://lists.linuxfromscratch.org/listinfo/lfs-support FAQ: http://www.linuxfromscratch.org/blfs/faq.html Unsubscribe: See the above information page Do not top post on this list. A: Because it messes up the order in which people normally read text. Q: Why is top-posting such a bad thing? A: Top-posting. Q: What is the most annoying thing in e-mail? http://en.wikipedia.org/wiki/Posting_style
