Skip to main content

Ralsina.Me — Roberto Alsina's website

About 80% of what's wrong with Linux users, all in a post

My post about Arch Lin­ux from yes­ter­day got post­ed at tux­ma­chines and there was a com­ment there.

I can't re­ply at the site be­cause:

  • It re­quires lo­­gin.

  • I can't find a place to get an ac­­coun­t.

  • It has freak­ing google ads pop­ups

So I re­ply here be­cause:

  • Hey, it's my own blog, so I can do what­ev­er I wan­t.

Here's the com­ment by hus­sam with my (ad­mit­ted­ly ranty) re­spon­se:

I've been us­ing Arch­Lin­ux as my on­ly dis­tri­bu­tion/­op­er­at­ing sys­tem since ear­ly 2006. It is re­al­ly a good dis­tri­bu­tion but late­ly there have been a lot of re­al­ly bad choic­es which I call bad com­pro­mis­es:

1. Too many Arch­Lin­ux users think gnome/kde are bloat and in­stead in­stall some half de­vel­oped win­dow man­ag­er and some ter­mi­nal em­u­la­tor and call it a "min­i­mal­ist" desk­top.

Why is that any of your busi­ness? And what "com­pro­mise" is there?

2. Op­tion­al de­pen­den­cies are the worst idea ev­er. If a pack­age is linked against lib­some­thing.­so then lib­some­thing should be a de­pen­den­cy not an op­tion­al de­pen­den­cy. Mak­ing lib­some­thing an op­tion­al de­pen­den­cy just be­cause "min­i­mal­ist" users don't want to in­stall de­pen­den­cies is plain stupid.

That's not what op­tion­al de­pen­den­cies are for. For ex­am­ple, con­sid­er the ex­am­ple I men­tioned, rst2pdf. It can use python­mag­ick. It can al­so not use it. You will lose one small fea­ture that AFAIK on­ly one per­son ev­er used. If you need that fea­ture, the man­u­al tells you what to do: in­stall python­mag­ick.

Maybe there should be a pac­man op­tion "in­stall opt­de­pend­s" which turns op­tion­al de­pen­den­cies in­to reg­u­lar ones. That would make you hap­py and keep oth­ers hap­py too.

3. Bad lead­er­ship. Aaron is fan­tas­tic guy but I know at least two Arch­Lin­ux de­vel­op­ers who can do a much bet­ter job.

That's just stupid and mean.

4. Too many Arch­Lin­ux users now like bad­ly au­to­ma­tion scripts like yaourt or what­ev­er it is called.

Parse er­ror. And then again: yaourt is great. You don't like it? Act as if it does­n't ex­ist and be hap­py. You seem to have a big prob­lem ig­nor­ing peo­ple who dis­agree with you. That's a re­al­ly, re­al­ly se­ri­ous per­son­al flaw. I sug­gest you grow up.

5. Too many noobs who do dumb things like peo­ple adding their users to hal, disk and dbus group­s.

Sure, they should add them­selves to op­ti­cal and stor­age. So what? It's a sim­ple prob­lem and it has a sim­ple so­lu­tion.

Then again, the ad­dus­er scripts prob­a­bly should do that for reg­u­lar us­er ac­counts. Af­ter al­l, who wants to cre­ate a reg­u­lar us­er that can't use re­mov­able stor­age? And if said use case ex­ist­s, that should be doable by re­mov­ing the user, and not vicev­er­sa!

On the oth­er hand, I don't give a damn, be­cause I can fix it triv­ial­ly.

The main rea­son why I don't think I will switch to an­oth­er dis­tri­bu­tion soon is that cre­at­ing Arch­lin­ux pack­ages from scratch is very easy and the initscript sys­tem is re­al­ly fan­tas­tic.

All in al­l, Arch­Lin­ux is a re­al­ly strong dis­tri­bu­tion now and it's con­stant­ly grow­ing.

I ex­pect you, like most elit­ist poseurs, will run away when you feel Arch is too pop­u­lar and ac­ces­si­ble to "too many noob­s" or some sim­i­lar non­sense.

Which, like the ti­tle says, is why you are a big part of what's wrong with Lin­ux user­s.

Why I STILL use Arch Linux

Yes­ter­day I had one of those mo­ments where I feel very hap­py about my dis­tro of choice, Arch Lin­ux. Since the last time I post­ed about Arch seems to have been over two years ago (time flies when you are hav­ing fun!), I think it's time to ex­plain it.

I want­ed to test rst2pdf against re­port­lab from SVN, wor­daxe from SVN and do­cu­tils from SVN, and I want­ed it to be sim­ple.

So­lu­tion: I just pack­aged them in AUR!

Now, whenever I need to check rst2pdf agains wordaxe trunk, I just need to yaourt -S python-­wor­dax­e-svn and I can go back to stable wordaxe with yaourt -S python-­wor­daxe.

The svn pack­age will al­ways be the cur­rent trunk with­out any mod­i­fi­ca­tion­s, and I can switch back and forth in about 45 sec­ond­s, with­out mess­ing up my sys­tem's pack­ages.

Also, I can keep my installed SVN packages updated by doing yaourt -Su --de­v­el every now and then.

How would I have done that us­ing De­bian or a RPM dis­tro? I sup­pose by go­ing around the pack­ag­ing sys­tem (which I hate) or by do­ing a pri­vate re­po (which is so ... lame?) or by do­ing a pub­lic re­po (which is freak­ing work).

Re­al­ly, if you are a coder, I can't think of a Lin­ux dis­tro that makes life eas­i­er than Arch. Pret­ty much ev­ery­thing is there (12K pack­ages in un­sup­port­ed!) and if it is­n't, it's a 5-minute job to slap it in­to AUR and help the com­mu­ni­ty.

Sup­pose you are do­ing a KDE ap­p. On most dis­tros you need to in­stall your own from-­source copy of kdelibs to have the lat­est and make sure it's not screwed by dis­tro-spe­cif­ic patch­es.

On Arch? Patch­ing up­stream is frowned up­on. Not hav­ing the lat­est ver­sion is frowned up­on. So it's pret­ty much the ide­al en­vi­ron­ment to de­vel­op against KDE, or GNOME, or PyQt or what­ev­er.

If my life was not 150% com­mit­ted al­ready, I would try to be­come an Arch de­vel­op­er, or at least a TU (Trust­ed User). Maybe next life!

Rst2pdf 0.11 released!

It's my plea­sure to an­nounce the re­lease of rst2pdf ver­sion 0.11, avail­able at http://­code.­google.­com/p/rst2pdf/­down­load­s/list

Rst2pdf is a tool to gen­er­ate PDF files di­rect­ly from re­struc­tured text sources via re­port­lab.

This ver­sion in­cludes many bug­fix­es and some new fea­tures com­pared to the pre­vi­ous 0.10.1 ver­sion, in­clud­ing but not lim­it­ed to em­bed­ding PDF im­ages, much im­proved im­age siz­ing, nicer list lay­out­s, bet­ter styling, page back­ground­s, and more than 15 bugs fixed. A full changel­og is in­clud­ed at the bot­tom of this mes­sage.

Rst2pdf aims to sup­port the full re­struc­tured text fea­ture set, and is very close to that goal, while al­so in­clud­ing some of the more ex­per­i­men­tal fea­tures, like a source code di­rec­tive with syn­tax high­light­ing and math no­ta­tion sup­port with La­TeX-­like syn­tax.

It sup­ports em­bed­ding ar­bi­trary fonts, both True Type and PS Type 1, both raster and vec­tor im­ages (in­clud­ing SVG), page tran­si­tion ef­fect­s, mul­ti­ple, flex­i­ble page lay­out­s, cas­cad­ing styles, and much, much more.

You can find more in­for­ma­tion about rst2pdf in its home page ( http://rst2pdf.­google­code.­com), and ask any­thing you want in the rst2pdf- dis­cuss mail­ing list (http://­group­s.­google.­com/­group/rst2pdf-dis­cuss)

Here are the changes in this ver­sion com­pared to 0.10.1:

  • De­­grade more grace­­ful­­ly when one or more wor­­daxe hy­phen­a­tors are bro­ken (cur­ren­t­­ly DWC is the bro­ken one)

  • Fixed is­­sue 132: in some cas­es, with user-de­fined fontAl­i­as, bold and ital­ic would get con­­fused (get­t­ing ital­ic in­­stead of bold in in­­­line markup, for in­­s­tance).

  • New stylesheet no-­­com­­pact-lists to make list­s... less com­­pact

  • SVG im­ages now han­­dle % as a width unit cor­rec­t­­ly.

  • Im­­ple­­men­t­ed is­­sue 127: sup­­port im­ages in PDF for­­mat. Right now they are ras­ter­ized, so it's not ide­al. Per­haps some­thing bet­ter will come up lat­er.

  • Fixed is­­sue 129: make it work around a prblem with Keep­­­To­geth­­er in RL 2.1 it prob­a­bly makes the out­­put look worse in some cas­es when us­ing that. RL 2.1 is not re­al­­ly sup­­port­ed, so added a warn­ing.

  • Fixed is­­sue 130: use os­­.­­path­sep in­­stead of ":" since ":" in win­­dows is used in disk names (and we still pay for DOS id­io­­cy, in 2009)

  • Fixed is­­sue 128: head­­ings lev­­el 3+ all looked the same

  • Ug­­ly bug­­fix for Is­­sue 126: crash­es when us­ing im­ages in head­­er + TOC

  • New tstyles sec­­tion in the stylesheet pro­­vides more con­­fig­urable list lay­outs and more pow­er­­ful ta­ble styling.

  • Bet­ter syn­­tax high­­­light­ing (sup­­ports bold/i­­tal­ic)

  • Work­around for is­­sue 103 so you can use bor­der­­Padding as a list (but it will look wrong if you are us­ing wor­­daxe <= 0.3.2)

  • Added field­­­val­ue style for field lists

  • Added op­­tion­list tstyle, for op­­tion lists

  • Added col­lec­­tion of util­i­­ty stylesheets and doc­u­­men­t­ed it

  • Im­proved com­­mand line pars­ing and stylesheet load­­ing (guess ex­ten­­sion like lat­est rst2la­­tex does)

  • Fixed Is­­sue 67: com­­plete­­ly new list lay­out­ing code

  • Fixed Is­­sue 116: crash­es caused by huge im­ages

  • Bet­ter sup­­port for %width in im­ages, n2ow it's % of the con­­tain­er frame's width, not of the text area.

  • Fixed bug in SVG scal­ing

  • Bet­ter han­dling of mis­s­ing im­ages

  • Added mis­s­ing styles ab­s­trac­t, con­­tents, ded­i­­ca­­tion to the de­­fault stylesheet

  • Ta­bles style sup­­port space­Be­­fore and spaceAfter

  • New top­ic-ti­­tle style for top­ic ti­­tles (ob­vi­ous ;-)

  • Ver­ti­­cal align­­ment for in­­­line im­ages (:align: pa­ram­e­ter)

  • Is­­sue 118: Sup­­port for :s­­cale: in im­ages and han­­dle re­­siz­ing of in­­­line im­ages

  • Is­­sue 119: Fix place­­ment of head­­ers and foot­ers

  • New back­­­ground prop­er­­ty for page tem­­plates (nice for pre­sen­­ta­­tion­s, for ex­am­­ple)

  • De­­fault to px for im­age width spec­i­­fi­­ca­­tions in­­stead of pt

  • Sup­­port all re­quired mea­­sure­­ment units ("em" "ex" "px" "in" "cm" "m­m" "p­t" "pc" "%" "")

  • New au­­to­­mat­ed scripts to check test cas­es for "vi­­su­al dif­fer­­ences"

  • Re­spect im­ages DPI prop­er­­ty a bit like rst2la­­tex does.

  • Is­­sue 110: New --in­­line-­­foot­notes op­­tion

  • Test­ed with re­­port­lab from SVN trunk

  • Sup­­port for Dinu Gher­­man's svglib. If both svglib and uni­­con­ver­­tor are avail­able, svglib is pre­­ferred (for SVG, of course). Patch orig­i­­nal­­ly by rute.

  • Is­­sue 109: Sep­a­rate styles for each kind of ad­­mo­ni­­tion

  • For Is­­sue 109: mis­s­ing styles are not a fa­­tal er­ror

  • Is­­sue 117: TOCs with more than 6 lev­­els now sup­­port­ed (raised lim­it to 9, which is sil­­ly deep)

Contents © 2000-2024 Roberto Alsina