/[gentoo]/xml/htdocs/proj/en/glep/glep-0033.txt
Gentoo

Diff of /xml/htdocs/proj/en/glep/glep-0033.txt

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

Revision 1.1 Revision 1.4
1GLEP: 33 1GLEP: 33
2Title: Eclass Restructure/Redesign 2Title: Eclass Restructure/Redesign
3Version: $Revision: 1.1 $ 3Version: $Revision: 1.4 $
4Last-Modified: $Date: 2005/02/16 21:30:44 $ 4Last-Modified: $Date: 2005/09/15 02:37:38 $
5Author: John Mylchreest <johnm@gentoo.org>, Brian Harring <ferringb@gentoo.org> 5Author: Brian Harring <ferringb@gentoo.org>, John Mylchreest <johnm@gentoo.org>
6Status: Draft 6Status: Draft
7Type: Standards Track 7Type: Standards Track
8Content-Type: text/x-rst 8Content-Type: text/x-rst
9Created: 29-Jan-2005 9Created: 29-Jan-2005
10Post-History: 29-Jan-2005 10Post-History: 29-Jan-2005 6-Mar-2005
11 11
12 12
13Abstract 13Abstract
14======== 14========
15 15
16For any design, the transition from theoretical to applied exposes inadequacies 16For any design, the transition from theoretical to applied exposes inadequacies
17in the original design. This document is intended to document, and propose a 17in the original design. This document is intended to document, and propose a
18revision of the current eclass setup to address current eclass inadequacies. 18revision of the current eclass setup to address current eclass inadequacies.
19 19
20This document proposes several thing- the creation of ebuild libraries, 'elibs', 20This document proposes several things- the creation of ebuild libraries, 'elibs',
21a narrowing of the focus of eclasses, a move of eclasses w/in the tree, the 21a narrowing of the focus of eclasses, a move of eclasses w/in the tree, the
22addition of changelogs, and a way to allow for simple eclass gpg signing. 22addition of changelogs, and a way to allow for simple eclass gpg signing.
23In general, a large scale restructuring of what eclasses are and how they're 23In general, a large scale restructuring of what eclasses are and how they're
24implemented. Essentially version two of the eclass setup. 24implemented. Essentially version two of the eclass setup.
25 25
27Terminology 27Terminology
28=========== 28===========
29 29
30From this point on, the proposed eclass setup will be called 'new eclasses', the 30From this point on, the proposed eclass setup will be called 'new eclasses', the
31existing crop (as of this writing) will be referenced as 'old eclasses'. The 31existing crop (as of this writing) will be referenced as 'old eclasses'. The
32destinction is elaborated on within this document. 32distinction is elaborated on within this document.
33 33
34 34
35Motivation and Rationale 35Motivation and Rationale
36======================== 36========================
37 37
38Eclasses within the tree currently are a bit of a mess- they're forced to 38Eclasses within the tree currently are a bit of a mess- they're forced to
39maintain backwards compatability w/ all previous functionality. In effect, 39maintain backwards compatibility w/ all previous functionality. In effect,
40their api is constant, and can only be added to- never changing the existing 40their api is constant, and can only be added to- never changing the existing
41functionality. This obviously is quite limiting, and leads to cruft accrueing in 41functionality. This obviously is quite limiting, and leads to cruft accruing in
42eclasses as a eclasses design is refined. This needs to be dealt with prior to 42eclasses as a eclasses design is refined. This needs to be dealt with prior to
43eclass code reaching a critical mass where they become unmanagable/fragile 43eclass code reaching a critical mass where they become unmanageable/fragile
44(recent pushes for eclass versioning could be interpretted as proof of this). 44(recent pushes for eclass versioning could be interpreted as proof of this).
45 45
46Beyond that, eclasses were originally intended as a method to allow for ebuilds 46Beyond that, eclasses were originally intended as a method to allow for ebuilds
47to use a pre-existing block of code, rather then having to duplicate the code in 47to use a pre-existing block of code, rather then having to duplicate the code in
48each ebuild. This is a good thing, but there are ill effects that result from 48each ebuild. This is a good thing, but there are ill effects that result from
49the current design. Eclasses inherit other eclasses to get a single function- in 49the current design. Eclasses inherit other eclasses to get a single function- in
50doing so, modifying the the exported 'template' (default src_compile, default 50doing so, modifying the the exported 'template' (default src_compile, default
51src_unpack, various vars, etc). All the eclass designer was after was reusing a 51src_unpack, various vars, etc). All the eclass designer was after was reusing a
52function, not making their eclass sensitive to changes in the template of the 52function, not making their eclass sensitive to changes in the template of the
53eclass it's inheriting. The eclass designer -should- be aware of changes in the 53eclass it's inheriting. The eclass designer -should- be aware of changes in the
54function they're using, but shouldn't have to worry about their default src_* 54function they're using, but shouldn't have to worry about their default src_*
55and pkg_* functions being overwritten, let alone the env changes. 55and pkg_* functions being overwritten, let alone the env changes.
56 56
61rather then requiring them to be aware of what phase of eclass changes is in 61rather then requiring them to be aware of what phase of eclass changes is in
62progress. 62progress.
63 63
64By rolling all changes into one large change, a line is intentionally drawn in 64By rolling all changes into one large change, a line is intentionally drawn in
65the sand. Old eclasses allowed for this, behaved this way. New eclasses allow 65the sand. Old eclasses allowed for this, behaved this way. New eclasses allow
66for that, and behave this way. This should reduce misconceptions about what is 66for that, and behave this way. This should reduce misconceptions about what is
67allowed/possible with eclasses, thus reducing bugs that result from said 67allowed/possible with eclasses, thus reducing bugs that result from said
68misconceptions. 68misconceptions.
69 69
70A few words on elibs- think of them as a clear definition between behavioral
71functionality of an eclass, and the library functionality. Eclass's modify
72template data, and are the basis for other ebuilds- elibs, however are *just*
73common bash functionality.
70 74
75Consider the majority of the portage bin/* scripts- these all are candidates for
76being added to the tree as elibs, as is the bulk of eutils.
77
78
71Specification. 79Specification
72============== 80=============
73 81
74The various parts of this proposal are broken down into a set of changes and 82The various parts of this proposal are broken down into a set of changes and
75elaborations on why a proposed change is preferable. It's advisable to the 83elaborations on why a proposed change is preferable. It's advisable to the
76reader that this be read serially, rather then jumping around. 84reader that this be read serially, rather then jumping around.
77 85
91A new directory named elib should be added to the top level of the tree to serve 99A new directory named elib should be added to the top level of the tree to serve
92as a repository of ebuild function libraries. Rather then relying on using the 100as a repository of ebuild function libraries. Rather then relying on using the
93source command, an 'elib' function should be added to portage to import that 101source command, an 'elib' function should be added to portage to import that
94libraries functionality. The reason for the indirection via the function is 102libraries functionality. The reason for the indirection via the function is
95mostly related to portage internals, but it does serve as an abstraction such 103mostly related to portage internals, but it does serve as an abstraction such
96that (for example) zsh compatability hacks could be hidden in the elib function. 104that (for example) zsh compatibility hacks could be hidden in the elib function.
97 105
98Elib's will be collections of bash functions- they're not allowed to do anything 106Elib's will be collections of bash functions- they're not allowed to do anything
99in the global scope aside from function definition, and any -minimal- 107in the global scope aside from function definition, and any -minimal-
100initialization of the library that is absolutely needed. Additionally, they 108initialization of the library that is absolutely needed. Additionally, they
101cannot modify any ebuild functions- src_compile, src_unpack fex. Since they are 109cannot modify any ebuild template functions- src_compile, src_unpack. Since they are
102required to not modify the metadata keys, nor in any way affect the ebuild aside 110required to not modify the metadata keys, nor in any way affect the ebuild aside
103from providing functionality, they can be conditionally pulled in. They also 111from providing functionality, they can be conditionally pulled in. They also
104are allowed to pull in other elibs, but strictly just elibs- no eclasses, just 112are allowed to pull in other elibs, but strictly just elibs- no eclasses, just
105other elibs. A realworld example would be the eutils eclass. 113other elibs. A real world example would be the eutils eclass.
106 114
107Portage, since the elib's don't modify metadata, isn't required to track elibs 115Portage, since the elib's don't modify metadata, isn't required to track elibs
108as it tracks eclasses. Thus a change in an elib doesn't result in half the tree 116as it tracks eclasses. Thus a change in an elib doesn't result in half the tree
109forced to be regenerated/marked stale when changed (this is more of an infra 117forced to be regenerated/marked stale when changed (this is more of an infra
110benefit, although regen's that take too long due to eclass changes have been 118benefit, although regen's that take too long due to eclass changes have been
111known to cause rsync issues due to missing timestamps). The only thing portage 119known to cause rsync issues due to missing timestamps).
112will do for elibs, aside from provide the elib function, is track what elibs 120
113have been loaded thus far, and load an elib only if it hasn't been loaded once 121Elibs will not be available in the global scope of an eclass, or ebuild- nor during the
114already. An implication of this (if it wasn't clear from the elib description) 122depends phase (basically a phase that sources the ebuild, to get its metadata). Elib
123calls in the global scope will be tracked, but the elib will not be loaded till just before
124the setup phase (pkg_setup). There are two reasons for this- first, it ensures elibs are
125completely incapable of modifying metadata. There is no room for confusion, late loading
126of elibs gives you the functionality for all phases, except for depends- depends being the
127only phase that is capable of specifying metadata. Second, as an added bonus, late
128loading reduces the amount of bash sourced for a regen- faster regens. This however is minor,
129and is an ancillary benefit of the first reason.
130
131There are a few further restrictions with elibs--mainly, elibs to load can only be specified
132in either global scope, or in the setup, unpack, compile, test, and install phases. You can
133not load elibs in prerm, postrm, preinst, and postinst. The reason being, for \*rm phases,
134installed pkgs will have to look to the tree for the elib, which allows for api drift to cause
135breakage. For \*inst phases, same thing, except the culprit is binpkgs.
136
115is that elibs cannot change their exported api dependant on the api (as some 137There is a final restriction--elibs cannot change their exported api dependent on the api
116eclass do for example). 138(as some eclass do for example). The reason mainly being that elibs are loaded once--not
139multiple times, as eclasses are.
140
141To clarify, for example this is invalid.
142::
143
144 if [[ -n ${SOME_VAR} ]]; then
145 func x() { echo "I'm accessible only via tweaking some var";}
146 else
147 func x() { echo "this is invalid, do not do it."; }
148 fi
149
117 150
118Regarding maintainability of elibs, it should be a less of a load then old 151Regarding maintainability of elibs, it should be a less of a load then old
119eclasses. One of the major issues with old eclasses is that their functions are 152eclasses. One of the major issues with old eclasses is that their functions are
120quite incestuous- they're bound tightly to the env they're defined in. This 153quite incestuous- they're bound tightly to the env they're defined in. This
121makes eclass functions a bit fragile- the restrictions on what can, and cannot 154makes eclass functions a bit fragile- the restrictions on what can, and cannot
122be done in elibs will address this, making functionality less fragile (thus a 155be done in elibs will address this, making functionality less fragile (thus a
123bit more maintainable). 156bit more maintainable).
124 157
125There is no need for backwards compatability with elibs- they just must work 158There is no need for backwards compatibility with elibs- they just must work
126against the current tree. Thus elibs can be removed when the tree no longer 159against the current tree. Thus elibs can be removed when the tree no longer
127needs them. The reasons for this are explained below. 160needs them. The reasons for this are explained below.
128 161
129Structuring of the elibs directory will be exactly the same as that of the new 162Structuring of the elibs directory will be exactly the same as that of the new
130eclass directory (detailed below), sans a different extension. 163eclass directory (detailed below), sans a different extension.
131 164
165As to why their are so many restrictions, the answer is simple- the definition of
166what elibs are, what they are capable of, and how to use them is nailed down as much as
167possible to avoid *any* ambiguity related to them. The intention is to make it clear,
168such that no misconceptions occur, resulting in bugs.
132 169
133The reduced role of Eclasses, and a clarification of existing Eclass requirements 170The reduced role of Eclasses, and a clarification of existing Eclass requirements
134--------------------------------------------------------------------------------- 171---------------------------------------------------------------------------------
135 172
136Since elibs are now intended on holding common bash functionality, the focus of 173Since elibs are now intended on holding common bash functionality, the focus of
137eclasses should be in defining an appropriate template for ebuilds. For example, 174eclasses should be in defining an appropriate template for ebuilds. For example,
138defining common DEPENDS, RDEPENDS, src_compile functions, src_unpack, etc. 175defining common DEPENDS, RDEPENDS, src_compile functions, src_unpack, etc.
139Additionally, eclasses should pull in any elibs they need for functionality. 176Additionally, eclasses should pull in any elibs they need for functionality.
140 177
141Eclass functionality that isn't directly related to the metadata, or src_* and 178Eclass functionality that isn't directly related to the metadata, or src_* and
142pkg_* funcs should be shifted into elibs to allow for maximal code reuse. This 179pkg_* funcs should be shifted into elibs to allow for maximal code reuse. This
151metadata keys exported from an ebuild on system A, must be *exactly* the same as 188metadata keys exported from an ebuild on system A, must be *exactly* the same as
152the keys exported on system B. 189the keys exported on system B.
153 190
154If an eclass (or ebuild for that matter) violates this constant requirement, it 191If an eclass (or ebuild for that matter) violates this constant requirement, it
155leads to portage doing the wrong thing for rsync users- for example, wrong deps 192leads to portage doing the wrong thing for rsync users- for example, wrong deps
156pulled in, leading to compilation failure. 193pulled in, leading to compilation failure, or dud deps.
157 194
158If the existing metadata isn't flexible enough for what is required for a 195If the existing metadata isn't flexible enough for what is required for a
159package, the parsing of the metadata is changed to address that. Cases where 196package, the parsing of the metadata is changed to address that. Cases where
160the constant requirement is violated are known, and a select few are allowed- 197the constant requirement is violated are known, and a select few are allowed-
161these are exceptions to the rule that are required due to inadequacies in 198these are exceptions to the rule that are required due to inadequacies in
162portage. In other words, those *few* exceptions are allowed because it's the 199portage. Any case where it's determined the constant requirement may need to be
163only way to do it at this time. Any case where it's determined the constant 200violated the dev must make it aware to the majority of devs, along with the portage
164requirement may need to be violated the dev must make it aware to the majority 201devs. This should be done prior to committing.
165of devs, and the portage devs- violation of the constant rule has far reaching
166effects.
167 202
168It's quite likely there is a way to allow what you're attempting- if you just go 203It's quite likely there is a way to allow what you're attempting- if you just go
169and do it, the rsync users (our userbase) suffer the results of compilation 204and do it, the rsync users (our user base) suffer the results of compilation
170failures and unneeded deps being pulled in. 205failures and unneeded deps being pulled in.
171 206
172After that stern reminder, back to new eclasses. Defining INHERITED and ECLASS 207After that stern reminder, back to new eclasses. Defining INHERITED and ECLASS
173within the eclass is no longer required. Portage already handles those vars if 208within the eclass is no longer required. Portage already handles those vars if
174they aren't defined. 209they aren't defined.
175 210
176As with elibs, it's no longer required backwards compatability be maintained 211As with elibs, it's no longer required that backwards compatibility be maintained
177indefinitely- compatability must be maintained against the current tree, but 212indefinitely- compatibility must be maintained against the current tree, but
178just that. As such new eclasses (the true distinction of new vs old is 213just that. As such new eclasses (the true distinction of new vs old is
179elaborated in the next section) can be removed from the tree once they're no 214elaborated in the next section) can be removed from the tree once they're no
180longer in use. 215longer in use.
181 216
182 217
183The end of backwards compatability... 218The end of backwards compatibility...
184------------------------------------- 219-------------------------------------
185 220
186With current eclasses, once the eclass is in use, it's api can no longer be 221With current eclasses, once the eclass is in use, its api can no longer be
187changed, nor can the eclass ever be removed from the tree. This is why we still 222changed, nor can the eclass ever be removed from the tree. This is why we still
188have *ancient* eclasses that are completely unused sitting in the tree, for 223have *ancient* eclasses that are completely unused sitting in the tree, for
189example inherit.eclass . The reason for this, not surprisingly is a portage 224example inherit.eclass. The reason for this, not surprisingly, is a portage
190deficiency- on unmerging an installed ebuild, portage used the eclass from the 225deficiency: on unmerging an installed ebuild, portage used the eclass from the
191current tree. 226current tree.
192 227
193For a real world example of this, if you merged a glibc 2 years back, whatever 228For a real world example of this, if you merged a glibc 2 years back, whatever
194eclasses it used must still be compatible, or you may not be able to unmerge the 229eclasses it used must still be compatible, or you may not be able to unmerge the
195older glibc version during an upgrade to a newer version. So either the glibc 230older glibc version during an upgrade to a newer version. So either the glibc
196maintainer is left with the option of leaving people using ancient versions out 231maintainer is left with the option of leaving people using ancient versions out
197in the rain, or maintaining an ever increasing load of backwards compatability 232in the rain, or maintaining an ever increasing load of backwards compatibility
198cruft in any used eclasses. 233cruft in any used eclasses.
199 234
200Binpkgs suffer a similar fate. Merging of a binpkg pulls needed eclasses from 235Binpkgs suffer a similar fate. Merging of a binpkg pulls needed eclasses from
201the tree, so you may not be able to even merge a binpkg if the eclasses api has 236the tree, so you may not be able to even merge a binpkg if the eclasses api has
202changed. If the eclass was removed, you can't even merge the binpkg, period. 237changed. If the eclass was removed, you can't even merge the binpkg, period.
205ebuild was built in already contains the eclasses functions, as such the env can 240ebuild was built in already contains the eclasses functions, as such the env can
206be re-used rather then relying on the eclass. In other words, binpkgs and 241be re-used rather then relying on the eclass. In other words, binpkgs and
207installed ebuilds will no longer go and pull needed eclasses from the tree, 242installed ebuilds will no longer go and pull needed eclasses from the tree,
208they'll use the 'saved' version of the eclass they were built/merged with. 243they'll use the 'saved' version of the eclass they were built/merged with.
209 244
210So the backwards compatability requirement for users of the next major portage 245So the backwards compatibility requirement for users of the next major portage
211version (and beyond) isn't required. All the cruft can be dropped. 246version (and beyond) isn't required. All the cruft can be dropped.
212 247
213The problem is that there will be users using older versions of portage that 248The problem is that there will be users using older versions of portage that don't
214don't support this functionality. So backwards compatability must be maintained 249support this functionality- these older installations *cannot* use the
215for them. Additionally, earlier versions of portage haven't always handled the 250new eclasses, due to the fact that their portage version is incapable of
216env correctly- for broken saved envs, the eclasses backwards compatability is 251properly relying on the env- in other words, the varying api of the eclass will
217still required. Waiting N months preserving backwards compatability in current 252result in user-visible failures during unmerging.
218eclasses, then dropping the support isn't much of an option. There always are
219stragglers who don't upgrade, beyond that, there is the possibility of cases
220where users -will- upgrade, but still be bitten (broken saved envs from earlier
221portage installations). More importantly, it doesn't provide a route to
222upgrade/fix things if a user lags behind, exempting trying to find a compatabile
223version of the eclass in viewcvs (assuming it hasn't been sent to the attic
224already). Obviously, that isn't acceptable.
225 253
226With the next major portage release, it will be possible to drop backwards 254So we're able to do a clean break of all old eclasses, and api cruft, but we need
227compatability for eclasses, and all lingering cruft. What is needed is a way to 255a means to basically disallow access to the new eclasses for all portage versions
228take full advantage of this functionality, without completely screwing over the 256incapable of properly handling the env requirements.
229unfortunates and those who don't upgrade.
230 257
231Unfortunately, the creation of new eclasses within the tree has an additional 258Unfortunately, we cannot just rely on a different grouping/naming convention within
232snag due to portage. The existing inherit function that is used to pull in old 259the old eclass directory. The new eclasses must be inaccessible, and portage throws
260a snag into this- the existing inherit function that is used to handle existing
233eclasses- basically, whatever it's passed (inherit kernel or inherit 261eclasses. Basically, whatever it's passed (inherit kernel or inherit
234kernel/kernel) it will pull in (kernel.eclass, and kernel/kernel.eclass 262kernel/kernel) it will pull in (kernel.eclass, and kernel/kernel.eclass
235respectively). So even if the new eclasses were implemented within a 263respectively). So even if the new eclasses were implemented within a
236subdirectory of the eclass dir in the tree, all current portage versions would 264subdirectory of the eclass dir in the tree, all current portage versions would
237still be able to access them. 265still be able to access them.
238 266
239In other words, these new eclasses would in effect, be old eclasses since older 267In other words, these new eclasses would in effect, be old eclasses since older
240portage versions could still access them. 268portage versions could still access them.
241 269
242 270
243Tree restructuring. 271Tree restructuring
244------------------- 272------------------
245 273
246There are only two way to block the existing (as of this writing) inherit 274There are only two way to block the existing (as of this writing) inherit
247functionality from accessing the new eclasses- either change the extension of 275functionality from accessing the new eclasses- either change the extension of
248eclasses to something other then 'eclass', or to have them stored in a seperate 276eclasses to something other then 'eclass', or to have them stored in a separate
249subdirectory of the tree then eclass. 277subdirectory of the tree then eclass.
250 278
251The latter is preferable, and the proposed solution. Reasons are- the current 279The latter is preferable, and the proposed solution. Reasons are- the current
252eclass directory is already overgrown. Structuring of the new eclass dir 280eclass directory is already overgrown. Structuring of the new eclass dir
253(clarified below) will allow for easier signing, ChangeLogs, and grouping of 281(clarified below) will allow for easier signing, ChangeLogs, and grouping of
257 285
258If it's unclear as to why the old inherit function *cannot* access the new 286If it's unclear as to why the old inherit function *cannot* access the new
259eclasses, please reread the previous section. It's unfortunately a requirement 287eclasses, please reread the previous section. It's unfortunately a requirement
260to take advantage of all that the next major portage release will allow. 288to take advantage of all that the next major portage release will allow.
261 289
262The proposed directory sructure is ${PORTDIR}/include/{eclass,elib}. 290The proposed directory structure is ${PORTDIR}/include/{eclass,elib}.
263Something like ${PORTDIR}/new-eclass, or ${PORTDIR}/eclass-ng could be used 291Something like ${PORTDIR}/new-eclass, or ${PORTDIR}/eclass-ng could be used
264(although many would cringe at the -ng), but such a name is unwise. Consider the 292(although many would cringe at the -ng), but such a name is unwise. Consider the
265possibility (likely a fact) that new eclasses someday may be found lacking, and 293possibility (likely a fact) that new eclasses someday may be found lacking, and
266refined further (version three as it were). Or perhaps we want to add yet more 294refined further (version three as it were). Or perhaps we want to add yet more
267functionality with direct relation to sourcing new files, and we would then need 295functionality with direct relation to sourcing new files, and we would then need
268to further populate ${PORTDIR}. 296to further populate ${PORTDIR}.
269 297
270The new-eclass directory will be (at least) 2 levels deep- for example: 298The new-eclass directory will be (at least) 2 levels deep- for example:
271 299
280 308
281No eclasses will be allowed in the base directory- grouping of new eclasses will 309No eclasses will be allowed in the base directory- grouping of new eclasses will
282be required to help keep things tidy, and for the following reasons. Grouping 310be required to help keep things tidy, and for the following reasons. Grouping
283of eclasses allows for the addition of ChangeLogs that are specific to that 311of eclasses allows for the addition of ChangeLogs that are specific to that
284group of eclasses, grouping of files/patches as needed, and allows for 312group of eclasses, grouping of files/patches as needed, and allows for
285saner/easier signing of eclasses- basically, you can just stick a signed 313saner/easier signing of eclasses- you can just stick a signed
286Manifest file w/in that grouping, thus providing the information portage needs 314Manifest file w/in that grouping, thus providing the information portage needs
287to ensure no files are missing, and that nothing has been tainted. 315to ensure no files are missing, and that nothing has been tainted.
288 316
289The elib directory will be structured in the same way, for the same reasons. 317The elib directory will be structured in the same way, for the same reasons.
290 318
291Repoman will have to be extended to work within new eclass and elib groups, and 319Repoman will have to be extended to work within new eclass and elib groups, and
292to handle signing and commiting. This is intentional, and a good thing. This 320to handle signing and committing. This is intentional, and a good thing. This
293gives repoman the possibility of doing sanity checks on elibs/new eclasses. 321gives repoman the possibility of doing sanity checks on elibs/new eclasses.
294It won't solve developers doing dumb things with eclasses (no technological
295solution would, exempting a tazering), but it will give us a way to automate
296checks to try and prevent honest mistakes from slipping through and breaking
297things for our users.
298 322
323Note these checks will not prevent developers from doing dumb things with eclass-
324these checks would only be capable of doing basic sanity checks, such as syntax checks.
325There is no way to prevent people from doing dumb things (exempting perhaps repeated
326applications of a cattle prod)- these are strictly automatic checks, akin to repoman's
327dependency checks.
299 328
329
300The start of a different phase of backwards compatability 330The start of a different phase of backwards compatibility
301--------------------------------------------------------- 331---------------------------------------------------------
302 332
303As clarified above, new eclasses will exist in a seperate directory that will be 333As clarified above, new eclasses will exist in a separate directory that will be
304intentionally inaccessible to the inherit function. As such, users of older 334intentionally inaccessible to the inherit function. As such, users of older
305portage versions *will* have to upgrade to merge any ebuild that uses elibs/new 335portage versions *will* have to upgrade to merge any ebuild that uses elibs/new
306eclasses. A depend on the next major portage version would address 336eclasses. A depend on the next major portage version would transparently handle
307transparently handle this for rsync users. 337this for rsync users.
308 338
309There still is the issue of users who haven't upgraded to the required portage 339There still is the issue of users who haven't upgraded to the required portage
310version. This is a minor concern frankly- portage releases include new 340version. This is a minor concern frankly- portage releases include new
311functionality, and bug fixes. If they won't upgrade, it's assumed they have 341functionality, and bug fixes. If they won't upgrade, it's assumed they have
312their reasons and are big boys, thus able to handle the complications themselves. 342their reasons and are big boys, thus able to handle the complications themselves.
313 343
314The real issue is broken envs, whether in binpkgs, or for installed packages. 344The real issue is broken envs, whether in binpkgs, or for installed packages.
315Two options exist- either the old eclasses are left in the tree indefinitely, or 345Two options exist- either the old eclasses are left in the tree indefinitely, or
329 359
330For users who do not upgrade within the window of N months while the old 360For users who do not upgrade within the window of N months while the old
331eclasses are in the tree, as stated, it's assumed they know what they are doing. 361eclasses are in the tree, as stated, it's assumed they know what they are doing.
332If they specifically block the new portage version, as the ebuilds in the tree 362If they specifically block the new portage version, as the ebuilds in the tree
333migrate to the new eclasses, they will have less and less ebuilds available to 363migrate to the new eclasses, they will have less and less ebuilds available to
334them. If they tried injecting the new portage version (lieing to portage, 364them. If they tried injecting the new portage version (lying to portage,
335essentially), portage would bail since it cannot find the new eclass. Note that 365essentially), portage would bail since it cannot find the new eclass.
336for them to even get to this point, they'd have to somehow disable the DEPEND on 366For ebuilds that use the new eclasses, there really isn't any way to sidestep
337a new version of portage- either hack up the ebuild, or do an injection. 367the portage version requirement- same as it has been for other portage features.
338Essentially they'd have to actively try to sidestep sanity checks implemented to
339make the shift over from old to new transparent. If they've
340disabled/sidestepped our attempt at a transparent migration, they can deal with
341the repercussions of it.
342 368
343What is a bit more annoying is that once the old eclasses are out of the tree, 369What is a bit more annoying is that once the old eclasses are out of the tree,
370if a user has not upgraded to a portage version supporting env processing, they
344users will lose the ability to unmerge any installed ebuild that used an old 371will lose the ability to unmerge any installed ebuild that used an old
345eclass, further users will lose the ability to merge any tbz2 that uses old 372eclass. Same cause, different symptom being they will lose the ability to merge
346eclasses. 373any tbz2 that uses old eclasses also.
347 374
348They however will *not* be left out in the rain. For merging old eclass 375There is one additional case that is a rarity, but should be noted- if a user
349binpkgs, and unmerging installed packages, they can merge the old eclass compat 376has suffered significant corruption of their installed package database (vdb). This is
350ebuild. The compat ebuild provides the missing eclasses, thus providing that 377ignoring the question of whether the vdb is even usable at this point, but the possibility
351lost functionality. 378exists for the saved envs to be non usable due to either A) missing, or B) corrupted.
379In such a case, even with the new portage capabilities, they would need
380the old eclass compat ebuild.
352 381
382Note for this to happen requires either rather... unwise uses of root, or significant
383fs corruption. Regardless of the cause, it's quite likely for this to even become an
384issue, the system's vdb is completely unusable. It's a moot issue at that point.
385If you lose your vdb, or it gets seriously damaged, it's akin to lobotomizing portage-
386it doesn't know what's installed, it doesn't know of its own files, and in general,
387a rebuilding of the system is about the only sane course of action. The missing env is
388truly the least of the users concern in such a case.
389
390Continuing with the more likely scenario, users unwilling to upgrade portage will
391*not* be left out in the rain. Merging the old eclass compat ebuild will provide
392the missing eclasses, thus providing that lost functionality.
393
353The intention isn't to force them to upgrade, hence the ability to restore the 394Note the intention isn't to force them to upgrade, hence the ability to restore the
354lost functionality. The intention is to clean up the existing mess, and allow us 395lost functionality. The intention is to clean up the existing mess, and allow us
355to move forward. The saying "you've got to break a few eggs to make an omelete" 396to move forward. The saying "you've got to break a few eggs to make an omelet"
356is akin, exempting the fact we're providing a way to make the eggs whole again 397is akin, exempting the fact we're providing a way to make the eggs whole again
357(the king's men would've loved such an option). 398(the king's men would've loved such an option).
358
359It's advisable that once all old eclasses are no longer in use in the tree, the
360old eclass package is added to system default. Remember that even those who
361have upgraded to a portage version that handles the env correctly, may run into
362instances where an installed packages env is corrupt. For new bootstraps (which
363automatically upgrade portage right off the bat), an injection of the compat
364package would be advisable- unless they downgrade portage, they will never need
365the old eclasses.
366 399
367 400
368Migrating to the new setup 401Migrating to the new setup
369-------------------------- 402--------------------------
370 403
384infrastructure server that generates the cache for rsync users will have to 417infrastructure server that generates the cache for rsync users will have to
385either be upgraded to a version of portage supporting new eclasses, or patched. 418either be upgraded to a version of portage supporting new eclasses, or patched.
386The former being much more preferable then the latter for the portage devs. 419The former being much more preferable then the latter for the portage devs.
387 420
388Beyond that, an appropriate window for old eclasses to exist in the tree must be 421Beyond that, an appropriate window for old eclasses to exist in the tree must be
389determined, and prior to that window passing an ebuild must be added to the tree 422determined, and prior to that window passing, an ebuild must be added to the tree
390so users can get the old eclasses if needed. 423so users can get the old eclasses if needed.
391 424
392For eclass devs to migrate from old to new, it is possible for them to just 425For eclass devs to migrate from old to new, it is possible for them to just
393transfer the old eclass into an appropriate grouping in the new eclass directory, 426transfer the old eclass into an appropriate grouping in the new eclass directory,
394although it's advisable they cleanse all cruft out of the eclass. You can 427although it's advisable they cleanse all cruft out of the eclass. You can
395migrate ebuilds gradually over to the new eclass, and don't have to worry about 428migrate ebuilds gradually over to the new eclass, and don't have to worry about
396having to support ebuilds from X years back. 429having to support ebuilds from X years back.
397 430
398Essentially, you have a chance to nail the design perfectly/cleanly, and have a 431Essentially, you have a chance to nail the design perfectly/cleanly, and have a
399window in which to redesign it. It's humbly suggested eclass devs take 432window in which to redesign it. It's humbly suggested eclass devs take
401 434
402 435
403Backwards Compatibility 436Backwards Compatibility
404======================= 437=======================
405 438
406All backwards compatability issues are addressed inline, but a recap is offered- 439All backwards compatibility issues are addressed in line, but a recap is offered-
407it's suggested that if the a particular compatability issue is 440it's suggested that if the a particular compatibility issue is
408questioned/worried over, the reader read the relevant section. There should be 441questioned/worried over, the reader read the relevant section. There should be
409a more in depth discussion of the issue, along with a more extensive explanation 442a more in depth discussion of the issue, along with a more extensive explanation
410of the potential solutions, and reasons for the choosen solution. 443of the potential solutions, and reasons for the chosen solution.
411 444
412To recap: 445To recap:
413:: 446::
414 447
415 New eclasses and elib functionality will be tied to a specific portage 448 New eclasses and elib functionality will be tied to a specific portage
416 version. A DEPENDs on said portage version should address this for rsync 449 version. A DEPENDs on said portage version should address this for rsync
417 users who refuse to upgrade to a portage version that supports the new 450 users who refuse to upgrade to a portage version that supports the new
418 eclasses/elibs and will gradually be unable to merge ebuilds that use said 451 eclasses/elibs and will gradually be unable to merge ebuilds that use said
419 functionality. It is their choice to upgrade, as such, the gradual 452 functionality. It is their choice to upgrade, as such, the gradual
420 'thinning' of available ebuilds should they block the portage upgrade is 453 'thinning' of available ebuilds should they block the portage upgrade is
421 their responsibility. 454 their responsibility.
422 455
423 Old eclasses at some point in the future should be removed from the tree, 456 Old eclasses at some point in the future should be removed from the tree,
424 and released in a tarball/ebuild. This will cause installed ebuilds that 457 and released in a tarball/ebuild. This will cause installed ebuilds that
425 rely on the old eclass to be unable to unmerge to behave as expected, with 458 rely on the old eclass to be unable to unmerge, with the same applying for
426 the same applying for merging of binpkgs. 459 merging of binpkgs dependent on the following paragraph.
427 460
428 This eclass ebuild should be a system depends target to make the transition 461 The old eclass-compat is only required for users who do not upgrade their
429 transparent. Future portage ebuilds, and the old eclass compat ebuild should 462 portage installation, and one further exemption- if the user has somehow
430 not inherit any eclasses. The reason for this is that in doing so, it may 463 corrupted/destroyed their installed pkgs database (/var/db/pkg currently),
431 block upgrade paths. At least for portage, this already is something of a 464 in the process, they've lost their saved environments. The eclass-compat
432 known issue for ebuild functionality- due to what it is/provides, it must 465 ebuild would be required for ebuilds that required older eclasses in such a
433 essentially be standalone, and cannot benefit from any eclass/elib 466 case. Note, this case is rare also- as clarified above, it's mentioned
434 functionality. 467 strictly to be complete, it's not much of a real world scenario as elaborated
468 above.
435 469
436 470
437Copyright 471Copyright
438========= 472=========
439 473

Legend:
Removed from v.1.1  
changed lines
  Added in v.1.4

  ViewVC Help
Powered by ViewVC 1.1.20