1 Oh Most Noble and Fragrant Emacs, please be in -*- outline -*- mode!
3 Newline Conversion and Keyword Substitution
4 ===========================================
8 We've finally settled on a proposal articulated by Greg Hudson,
9 derived from Ben's original proposal plus much list discussion.
11 Here's Greg's mail with the proposal, then a few clarifying selections
12 from the followup discussion:
14 Alright, I'll make a proposal which is like yours but (in my
15 opinion) a little clearer. First, let's look at the different use
18 1. The most common case--text files which want native line endings.
19 These should be stored in the repository using LF line endings,
20 and in the working dir using native line endings.
22 2. Binary files. These files we don't want to touch at all.
24 3. Text files which, for one reason or another, want a specific
25 line ending format regardless of platform. These should be
26 stored in the repository and in the working directory using the
27 specified line ending. We probably don't have to worry so much
28 about data safety for these files since a particular, odd
29 behavior has been specified for them.
31 There are, of course, a hundred different ways we could arrange the
32 metadata. I propose an "svn:newline-style" property with the
33 possible values "none", "native", "LF", "CR", and "CRLF". The
36 none: Use case 2. don't do any newline translation
38 native: Use case 1. Store with LF in repository, and with native
39 line endings in the working copy.
41 LF, CR, CRLF: Use case 3. Store with specified format in the
42 repository and in the working copy.
44 On commit, we apply the following rules to transform the data
45 committed to the server:
47 If newline-style is none, do nothing.
49 If newline-stle is native, translate <native newline style> -> LF.
50 If we notice any CRs or LFs which aren't part of a native-style
51 newline, abort the commit.
53 If newline-style is LF, CR, or CRLF, translate <native newline
54 style> -> <requested newline style>. If we notice any CRs or LFs
55 which aren't part of a native-style newline and aren't part of a
56 requested-style newline, abort the commit. If the commit succeeds,
57 apply the <native newline style> -> <requested newline style>
58 translation to the working copy as well, so that it matches what we
59 would get from a checkout of the new rev.
61 On checkout, we translate LF -> <native newline style> if
62 newline-style is native; otherwise, we leave the file alone.
64 For now, let's say the default value of svn:newline-style is none.
65 In the future, we'll want to think about things like how to enable
66 newline-translation over the whole repository except for files
67 which don't appear to be text.
69 I think that's a complete proposal. Some possible variations:
71 Variation 1: If newline-style is native, on commit, translate
72 <first newline style seen> -> LF. If we see any CRs or LFs which
73 don't match the first newline style seen, abort the commit.
75 Variation 2: If newline-style is native, before commit, examine the
76 file to see if it uses only the native newline style. If it
77 doesn't, set the newline-style property to "none" and commit with
80 Variation 3: Combine variations 1 and 2; if newline-style is
81 native, then if before commit, examine the file to see if it uses a
82 single consistent newline style. If it does, translate <that
83 newline style> -> LF; if not, commit with newline-style set to
84 "none" and no translation.
86 Variation 4: If newline-style is native, then on commit, we edit a
87 property "svn:newline-conversion" to something like "CRLF LF" to
88 show what conversion we did. This enables mechanical reversal of
89 the translation if the file is later determined to be binary.
90 (Particularly useful with variations 1 or 3 where the transform
91 might not be obvious from the platform where the file was checked
94 We decided to hold off on doing any of the "variations". We'll just
95 do the basic proposal first, then see how well things work out.
97 Next, Greg responded to an observation by William Uther, in which
98 William pointed out that the behavior is not reversible when
99 svn:newline-style is LF or CRLF. Greg agrees, but explains why that's
102 On Fri, 2001-12-14 at 15:48, William Uther wrote:
103 > --On Friday, 14 December 2001 1:16 PM -0500 Greg Hudson
104 > <ghudson@MIT.EDU> wrote:
106 > > If newline-style is LF, CR, or CRLF, translate <native
107 > > newline style> -> <requested newline style>. If we notice any
108 > > CRs or LFs which aren't part of a native-style newline and
109 > > aren't part of a requested-style newline, abort the commit. If
110 > > the commit succeeds, apply the <native newline style> ->
111 > > <requested newline style> translation to the working copy as
112 > > well, so that it matches what we would get from a checkout of
115 > I don't think this preserves reversability. If a file contains
116 > BOTH <native-style newline> and <requested-style newline> then
117 > you neet to abort. If you translate just <native-style newline>
118 > then you can't undo the transformation - you don't know which
119 > newlines need to be untransformed.
121 This particular transform (for files marked CRLF, CR, or LF) is not
122 reversible. See where I said:
124 "We probably don't have to worry so much about data safety for
125 these files since a particular, odd behavior has been specified
128 However, let's add a possible variation to my proposal, for those
129 who are still uncomfortable with data-destroying transformations
130 applied to such flies:
132 Variation 5: If the file is marked CRLF, CR, or LF, we translate
133 <native-style newline> to <requested-style newline> during commit,
134 and abort the commit if we notice any kind of mixing of newline
135 styles. (Can also combine with variation 1.)
137 Colin Putney also followed up to William Uther's post, agreeing with
138 Greg and explaining why in even more detail:
140 > I don't think this preserves reversability. If a file contains
141 > BOTH <native-style newline> and <requested-style newline> then
142 > you neet to abort. If you translate just <native-style newline>
143 > then you can't undo the transformation - you don't know which
144 > newlines need to be untransformed.
146 > Stated simply: You should only translate when the newline style
147 > is entirely consistent. Anything else removes the inconsistency
148 > and hence loses information.
150 True, this scheme doesn't preserve reversibility. But in this case
151 that's OK, because the newline-style decrees what the newline style
152 must be. If there are native-style newlines mixed in with the
153 requested-style newlines, this is probably the result of corruption
154 by some native-newline-obsessive user tool. So the non-reversible
155 transform will actually undo the corruption.
157 For example, the file foo.dsp, which has newline-style of
158 CRLF. It's stored in the repository with CRLF newlines and on
159 checkout, no transformation is done. If Linus checks out the file
160 and edits it in an old version of emacs, any lines he adds will be
161 terminated with a bare LF. Since this is his native style of
162 newline, the transformation Greg described will undo this damage.
164 If the newline-style is set to a specific newline-style (ie. CR,
165 LF, or CRLF), then we know that (1) the file is text, not binary,
166 and (2), any other style of newline present is corruption.
168 A file should not be marked with a specific newline style unless
169 (1) user does so explicitly, or (2) it matches some heuristic when
170 it's added, *and* the file contents conform to that newline style.
172 So the only real possibility for corruption is if some user tool
173 creates a binary file that matches a heuristic for a specific
174 newline style. In our running example, William creates a vector
175 graphics file called foo.dsp and adds it. By chance, this file
176 happens to have CRLFs scattered though it, but no bare CRs, LFs,
177 '\0' characters or other harbingers of binary files. On the commit,
178 svn will notice the extension, set the newline-style to CRLF and
179 send it to the repository. William may get an error if he tries to
180 commit a change that introduces a bare CR or LF, but he won't
183 Linus can corrupt the file if he makes a change that introduces a
184 bare LF, which will get transformed into CRLF on
185 commit. Alternatively, Madeleine (was that her name?) Can introduce
186 a bare CR and commit, which will also corrupt the file.
188 That's a pretty long string of unlikely coincidences though, while
189 the opposite case, where this transformation *fixes* corruption, is
194 Finally, Greg followed up, confirming again what he meant by that
195 portion of the proposal, but also saying that he could go the other
196 way on the question too:
198 > +1 on Greg Hudson's latest proposal -- and I think we're now
199 > ready to Actually Do It. :-)
201 I hope so. For a while I was afraid we had hit our first failure
202 to achieve livable consensus. My apologies for not realizing the
203 reversability thing until two days and several thousand lines of
204 misguided debate had already gone by.
206 > My assumption is that "in the working copy" means both text-base
207 > and working file, for the sake of an efficient is-modified-p
208 > test, and since the repository file is just an automatic
209 > transform off the text-base anyway.
211 Actually, I was assuming that text-base would be a verbatim copy of
212 the repository contents. But that's kind of an implementation
213 detail; let's leave that up to Ben (assuming he's doing the
216 > Otherwise, then the is-modified-p check has to be tweaked in a
217 > way that will make modifiedness checks a lot slower in some
220 No... it just means that if the mod times force a contents check,
221 you have to translate the text-base contents as you compare them
222 against the normal contents. That's "a teeny tiny bit slower,"
225 > The second sentence of the above paragraph isn't about allowing
226 > mixed-style files. It's saying that if the entire file is native
227 > format, allow that (and transform when necessary), OR if the
228 > entire file is in the requested style, then allow that too. The
229 > latter situation could happen if someone used a LF-style tool
230 > under Windows, for example, so that when an LF-style file got
231 > saved, the whole thing would be LF-style now, not native style.
232 > No reason to disallow this.
236 See my last message, as well as Colin Putney's argument. In
237 summary, that's not actually what I meant, but I don't really care
240 We're implementing Greg's original meaning, since (as Colin Putney
241 pointed out) the chance of "bad" data corruption is very small, and
242 the chance that it would undo corruption is actually greater.
245 * Keyword substitution
247 Quick summary: there's one property, named "svn:keywords". It's value
248 is a whitespace-separated list of keywords to expand. Since keywords
249 have both long and short forms, both ways are allowed in the value of
250 the svn:keywords property. Here are all the keywords:
252 "LastChangedBy" "Author" ---> either one expands author
253 "LastChangedDate" "Date" ---> either one expands date
254 "LastChangedRevision" "Revision" "Rev" ---> any one expands rev
255 "HeadURL" "URL" ---> either one expands url
257 Here are some example values of the property:
259 "Rev LastChangedDate HeadURL"
261 "Author\nDate \n LastChangedRevision URL"
263 Unrecognized words are ignored; absence of the property is the same as
264 an empty value or a value with no valid keywords in it.
266 Keywords (long and short forms) are case-sensitive, as in CVS.