# Understanding SHAPE and MOLD **URL:** \ **Category:** Help\ **Created:** [December 30, 2024, 3:04pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018 "2024-12-30T15:04:34Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![bosse](https://avatars.discourse-cdn.com/v4/letter/b/d07c76/32.png) [@bosse](https://fortran-lang.discourse.group/u/bosse)\ **Post date:** [December 30, 2024, 3:04pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/1 "2024-12-30T15:04:34Z")
I am new to this group but I have programmed in Fortran since 1977. The new Fortran has been a great help in creating the fairly large software for thermodynamic calculation but recently I have come across problem with a very simple use of allocation. I am sorry I do not know how to format the code nicely below. > Program allosp > implicit none > integer i > type species > character(len=:) , allocatable :: name > integer nel > character(len=:), dimension(:), allocatable :: element > double precision, dimension(:), allocatable :: stoic > double precision charge > end type species > ! type(species), dimension(:), allocatable :: const > type(species), dimension(10), target :: const > type(species), pointer :: sp > ! > ! allocate(const(10)) > sp=\>const(1) > sp%name=‘H2O’ > sp%nel=2 > allocate(sp%element(2)) > allocate(sp%stoic(1:2)) > end Program allosp Compiling this gives Error: Allocate-object at (1) with a deferred type parameter requires either a type-spec or SOURCE tag or a MOLD tag Initially I thought that it was a problem to allocate something inside an allocatable object but that is not the case. During the last 15 years or so using the new Fortran I have never had this kind of error and I have used allocate many times. In my old code I find statements as: ``` allocate(ceq%complist(icomp)%endmember(nsl)) ``` where ceq is a very complex object with many allocatable TYPE objects for all the data needed to describe the equilibrium state of a material. Maybe I had some trick 15 years ago that I forgotten now. Or is it something new? But my old code still compiles correctly. Any hint welcome Bosse ---
**Author:** ![themos](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/themos/32/521_2.png) [@themos](https://fortran-lang.discourse.group/u/themos)\ **Post date:** [December 30, 2024, 4:04pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/2 "2024-12-30T16:04:10Z")
_Deferred type parameters_ appear in your program for components NAME and ELEMENT. For NAME, the LEN type parameter for CHARACTER is specified by the r.h.s. ‘H2O’, to be 3. The > [@bosse](#): > > allocate(sp%element(2)) similarly requires a specification for the length type parameter, which can be done with a type-spec ```auto allocate(CHARACTER(LEN=3) :: sp%element(2)) ``` ,MOLD ```auto allocate(sp%element(2),MOLD='H2O') ``` or SOURCE specifier ```auto allocate(sp%element(2),SOURCE='H2O') ``` ---
**Author:** ![jwmwalrus](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/jwmwalrus/32/4483_2.png) [@jwmwalrus](https://fortran-lang.discourse.group/u/jwmwalrus)\ **Post date:** [December 30, 2024, 4:09pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/3 "2024-12-30T16:09:55Z")
Welcome to the forum. When you declared the component ```fortran character(len=:), dimension(:), allocatable :: element ``` You stated that it has both a deferred **length** and a deferred **shape**. But when doing the allocation ```fortran allocate(sp%element(2)) ``` You’re only providing a **shape** (i.e., the size). You need to do something like ```fortran allocate(character(255) :: sp%element(2)) ``` Btw, to properly format code in posts, you can either use markdown or the `` icon from the toolbar. For markdown, it should be something like this: ````auto ```fortran code goes here ``` ```` ---
**Author:** ![bosse](https://avatars.discourse-cdn.com/v4/letter/b/d07c76/32.png) [@bosse](https://fortran-lang.discourse.group/u/bosse)\ **Post date:** [December 30, 2024, 5:32pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/4 "2024-12-30T17:32:56Z")
Thanks for the answers, I am not used to allocatable characters, I do not think they existed when I started writing in the new Fortran standard around 2010. However, recently I got used to just define characters as allocatable and when they are assigned a text they are automatically allocated. But I understand combining dimension and len= makes it a bit complicated but I find this MOLD very strange. I think I have picked up somewhere that in a character array all elements must have the same length. ---
**Author:** ![jwmwalrus](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/jwmwalrus/32/4483_2.png) [@jwmwalrus](https://fortran-lang.discourse.group/u/jwmwalrus)\ **Post date:** [December 30, 2024, 6:02pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/5 "2024-12-30T18:02:21Z")
Deferred-length characters were introduced in Fortran 2003. They’re proper strings, but the `len` parameter belongs to the variable and not to the elements, so deferred-length character arrays have that limitation. The reallocate-on-assignment feature applies to any `allocatable` variable (not only to deferred-length characters)… and the `allocate` attribute can be used on scalars as well. As for `MOLD=`, it copies type parameters during allocation (e.g., length and dimension), but not content. `SOURCE=` copies the whole thing, but there might be issues with partially allocated variables —you get the same effect with reallocate-on-assignment, with the added benefit that partial allocation is not an issue. ---
**Author:** ![themos](https://yyz2.discourse-cdn.com/free1/user_avatar/fortran-lang.discourse.group/themos/32/521_2.png) [@themos](https://fortran-lang.discourse.group/u/themos)\ **Post date:** [December 30, 2024, 6:56pm UTC](https://fortran-lang.discourse.group/t/understanding-shape-and-mold/9018/6 "2024-12-30T18:56:36Z")
You can only have one of MOLD/SOURCE= specifiers. SOURCE=_expr_ will use the value of _expr_ to define the object that is being allocated. MOLD=_expr_ will not use the value of _expr_ (which indeed can be undefined) and will **not** define any values for the object being allocated. It is likely to be faster. The _expr_ in that case is only used for its type, type parameters and attributes.