Page 33

The back­quoted com­mand wc -l /etc/passwd seen in the page's last com­mand line should ac­tu­ally be wc -l < /etc/passwd.

The issue is that wc -l prints out file­names as well as line counts when file­names are sup­plied on the com­mand line. Here, only the line count is wanted.

Page 44

Con­trary to Table 2.2, bash has no string equiv­a­lents to the nu­meric -le and -ge op­er­a­tors. The >= and <= op­er­a­tors listed in Table 2.2 do not exist.

The > and < op­er­a­tors do exist. How­ever, they must be back­slash-es­caped or dou­ble brack­eted to pre­vent them from being in­ter­preted as I/O redi­rec­tion char­ac­ters.

Page 45

The loop vari­able in the ex­am­ple for loop is ac­tu­ally called script. The text refers to it as file.

Pages 51 and 52

The reg­u­lar ex­pres­sion shown on pages 51–52 for match­ing var­i­ous spellings of Moam­mar Gad­hafi's name is OK as orig­i­nally printed. How­ever, the ac­com­pa­ny­ing text does not ex­plain why the ([dtz][dhz]?)+ group (a com­po­nent of the last name match) al­lows rep­e­ti­tion.

The + is there to ac­com­mo­date vari­ants such as Quad­hd­hafi. Re­ally, that frag­ment should be ([dtz][dhz]?){1,2} so that only one or two oc­cur­rences are per­mis­si­ble.

More gen­er­ally, it's worth not­ing that this regex is per­mis­sive. Not every string that it matches is a le­git­i­mate spelling of Gad­hafi's name. How­ever, it should match all le­git­i­mate spellings.

Page 153

The cross-ref­er­ence in the last line of the first para­graph under The se­tuid and set­gid bits points to the wrong place.

The ref­er­enced ma­te­r­ial is on page 105, not page 106.

Page 466

The route com­mands at the bot­tom of the page are shell com­mands, not ex­cerpts from a startup script as the sur­round­ing text might imply.

The # char­ac­ter in­di­cates root's shell prompt, not the com­ment char­ac­ter used in shell scripts.

Page 574

DNS Pun­y­code names begin with xn--, not xf-- as stated in the text.

Page 808

In ad­di­tion to hav­ing its own con­ven­tions, De­bian's ver­sion of Exim con­fig­u­ra­tion uses the m4 pre­proces­sor.

We don't de­scribe the De­bian sys­tem in this book.

Page 817

There are two er­rors on this page.

In the sen­tence "The sender's ad­dress is not checked be­cause many user agents get con­fused by error re­turns," re­place "sender's ad­dress" with "re­cip­i­ent's ad­dress."

The para­graph that starts "We next check the des­ti­na­tion do­main..." claims that DKIM ver­i­fi­ca­tion is not per­formed dur­ing this step. That is in­cor­rect; DKIM is in fact en­abled.

Page 821

It is not en­tirely ac­cu­rate to say that "If a mes­sage re­ceives a pass or decline from all the routers in the se­quence, it is bounced as an un­routable ad­dress."

In fact, Exim may ei­ther bounce or re­ject such mes­sages, de­pend­ing on the con­text.

Page 827

Con­trary to the state­ment in the De­bug­ging sec­tion, exim does fork when given the -d flag. How­ever, it does not de­tach it­self from the con­trol ter­mi­nal.

Page 852

In the con­fig­u­ra­tion file ex­cerpt, the line that sets dkim_selector should do a lookup on $dkim_domain, not $sender_address_domain as shown.